2022-12-07 04:36:17,368 INFO [train.py:941] (1/4) Training started 2022-12-07 04:36:17,368 INFO [train.py:951] (1/4) Device: cuda:1 2022-12-07 04:36:17,415 INFO [lexicon.py:168] (1/4) Loading pre-compiled data/lang_char/Linv.pt 2022-12-07 04:36:17,423 INFO [train.py:962] (1/4) {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 100, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'b2ce63f3940018e7b433c43fd802fc50ab006a76', 'k2-git-date': 'Wed Nov 23 08:43:43 2022', 'lhotse-version': '1.9.0.dev+git.97bf4b0.dirty', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'ali_meeting', 'icefall-git-sha1': 'f13cf61-dirty', 'icefall-git-date': 'Tue Dec 6 03:34:27 2022', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r8n04', 'IP address': '10.1.8.4'}, 'world_size': 4, 'master_port': 12354, 'tensorboard': True, 'num_epochs': 20, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp/v1'), 'lang_dir': 'data/lang_char', 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'context_size': 2, 'prune_range': 5, 'lm_scale': 0.25, 'am_scale': 0.0, 'simple_loss_scale': 0.5, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 5000, 'keep_last_k': 10, 'average_period': 200, 'use_fp16': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'manifest_dir': PosixPath('data/manifests'), 'enable_musan': True, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'max_duration': 300, 'max_cuts': 100, 'num_buckets': 50, 'on_the_fly_feats': False, 'shuffle': True, 'num_workers': 8, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'blank_id': 0, 'vocab_size': 3290} 2022-12-07 04:36:17,424 INFO [train.py:964] (1/4) About to create model 2022-12-07 04:36:17,814 INFO [zipformer.py:179] (1/4) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2022-12-07 04:36:17,861 INFO [train.py:968] (1/4) Number of model parameters: 75734561 2022-12-07 04:36:22,523 INFO [train.py:983] (1/4) Using DDP 2022-12-07 04:36:23,012 INFO [asr_datamodule.py:357] (1/4) About to get AMI train cuts 2022-12-07 04:36:23,014 INFO [asr_datamodule.py:204] (1/4) About to get Musan cuts 2022-12-07 04:36:23,014 INFO [asr_datamodule.py:208] (1/4) Enable MUSAN 2022-12-07 04:36:24,269 INFO [asr_datamodule.py:232] (1/4) Enable SpecAugment 2022-12-07 04:36:24,269 INFO [asr_datamodule.py:233] (1/4) Time warp factor: 80 2022-12-07 04:36:24,269 INFO [asr_datamodule.py:246] (1/4) About to create train dataset 2022-12-07 04:36:24,269 INFO [asr_datamodule.py:259] (1/4) Using DynamicBucketingSampler. 2022-12-07 04:36:24,630 INFO [asr_datamodule.py:268] (1/4) About to create train dataloader 2022-12-07 04:36:24,631 INFO [asr_datamodule.py:381] (1/4) About to get AliMeeting IHM eval cuts 2022-12-07 04:36:24,632 INFO [asr_datamodule.py:300] (1/4) About to create dev dataset 2022-12-07 04:36:24,827 INFO [asr_datamodule.py:315] (1/4) About to create dev dataloader 2022-12-07 04:36:55,256 INFO [train.py:873] (1/4) Epoch 1, batch 0, loss[loss=5.146, simple_loss=4.666, pruned_loss=4.789, over 13501.00 frames. ], tot_loss[loss=5.146, simple_loss=4.666, pruned_loss=4.789, over 13501.00 frames. ], batch size: 17, lr: 2.50e-02, grad_scale: 2.0 2022-12-07 04:36:55,256 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 04:37:02,329 INFO [train.py:905] (1/4) Epoch 1, validation: loss=4.832, simple_loss=4.375, pruned_loss=4.552, over 857387.00 frames. 2022-12-07 04:37:02,329 INFO [train.py:906] (1/4) Maximum memory allocated so far is 15430MB 2022-12-07 04:37:05,274 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:37:05,405 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=87.80 vs. limit=5.0 2022-12-07 04:37:10,012 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=7.05 vs. limit=2.0 2022-12-07 04:37:18,878 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:37:32,955 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=198.62 vs. limit=5.0 2022-12-07 04:37:40,079 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0594, 4.0593, 4.0561, 4.0503, 3.4719, 3.9903, 4.0148, 4.0192], device='cuda:1'), covar=tensor([0.0085, 0.0050, 0.0039, 0.0052, 0.0065, 0.0064, 0.0043, 0.0045], device='cuda:1'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:1'), out_proj_covar=tensor([9.1956e-06, 9.2484e-06, 9.0974e-06, 9.3481e-06, 9.2369e-06, 9.2645e-06, 9.1997e-06, 9.3760e-06], device='cuda:1') 2022-12-07 04:37:43,999 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=22.82 vs. limit=2.0 2022-12-07 04:37:44,447 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=6.88 vs. limit=2.0 2022-12-07 04:37:48,807 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=147.64 vs. limit=5.0 2022-12-07 04:38:01,922 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:38:03,502 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=16.03 vs. limit=2.0 2022-12-07 04:38:13,697 INFO [train.py:873] (1/4) Epoch 1, batch 100, loss[loss=0.4526, simple_loss=0.4063, pruned_loss=0.3888, over 2547.00 frames. ], tot_loss[loss=0.972, simple_loss=0.8754, pruned_loss=0.8866, over 904288.08 frames. ], batch size: 100, lr: 3.00e-02, grad_scale: 0.125 2022-12-07 04:38:16,996 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 3.426e+01 1.096e+02 2.039e+02 4.080e+02 7.158e+03, threshold=4.078e+02, percent-clipped=0.0 2022-12-07 04:38:28,320 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=8.83 vs. limit=2.0 2022-12-07 04:38:34,195 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=5.73 vs. limit=2.0 2022-12-07 04:38:42,236 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:38:50,175 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6722, 3.6146, 3.8006, 3.7621, 3.6831, 3.7761, 3.7938, 3.7849], device='cuda:1'), covar=tensor([0.0046, 0.0028, 0.0019, 0.0038, 0.0036, 0.0029, 0.0023, 0.0029], device='cuda:1'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:1'), out_proj_covar=tensor([8.8579e-06, 8.8805e-06, 8.7757e-06, 8.9963e-06, 8.8975e-06, 8.9016e-06, 8.8526e-06, 9.0117e-06], device='cuda:1') 2022-12-07 04:39:03,242 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8418, 3.8361, 3.8397, 3.8366, 3.8456, 3.8358, 3.8321, 3.8244], device='cuda:1'), covar=tensor([0.0024, 0.0029, 0.0022, 0.0021, 0.0015, 0.0031, 0.0038, 0.0025], device='cuda:1'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:1'), out_proj_covar=tensor([8.9646e-06, 8.9072e-06, 9.1700e-06, 8.9149e-06, 9.1003e-06, 9.0624e-06, 9.0270e-06, 9.0327e-06], device='cuda:1') 2022-12-07 04:39:21,526 INFO [train.py:873] (1/4) Epoch 1, batch 200, loss[loss=0.5012, simple_loss=0.4249, pruned_loss=0.5126, over 13890.00 frames. ], tot_loss[loss=0.6972, simple_loss=0.6157, pruned_loss=0.6704, over 1318014.09 frames. ], batch size: 23, lr: 3.50e-02, grad_scale: 0.25 2022-12-07 04:39:24,001 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.49 vs. limit=2.0 2022-12-07 04:39:24,206 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0915, 4.0919, 4.0907, 4.0887, 4.0912, 4.0922, 4.0902, 4.0922], device='cuda:1'), covar=tensor([1.0274e-04, 1.6038e-04, 6.1217e-05, 1.4225e-04, 9.7870e-05, 1.0264e-04, 1.1656e-04, 1.4891e-04], device='cuda:1'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:1'), out_proj_covar=tensor([9.2883e-06, 9.3747e-06, 9.2683e-06, 8.9862e-06, 9.3520e-06, 9.1465e-06, 9.2779e-06, 9.1566e-06], device='cuda:1') 2022-12-07 04:39:24,801 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 2.748e+01 6.194e+01 9.239e+01 1.604e+02 3.274e+02, threshold=1.848e+02, percent-clipped=0.0 2022-12-07 04:40:10,977 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=5.06 vs. limit=2.0 2022-12-07 04:40:22,373 WARNING [optim.py:389] (1/4) Scaling gradients by 0.06466581672430038, model_norm_threshold=184.7753143310547 2022-12-07 04:40:22,532 INFO [optim.py:451] (1/4) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.6.weight with proportion 0.54, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=4.438e+06, grad_sumsq = 3.527e+09, orig_rms_sq=1.258e-03 2022-12-07 04:40:27,561 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:40:30,373 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:40:31,031 INFO [train.py:873] (1/4) Epoch 1, batch 300, loss[loss=0.4874, simple_loss=0.412, pruned_loss=0.4518, over 14227.00 frames. ], tot_loss[loss=0.5851, simple_loss=0.5094, pruned_loss=0.5588, over 1579341.74 frames. ], batch size: 60, lr: 4.00e-02, grad_scale: 0.5 2022-12-07 04:40:34,362 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 2.885e+01 5.232e+01 7.085e+01 9.806e+01 2.857e+03, threshold=1.417e+02, percent-clipped=2.0 2022-12-07 04:40:40,446 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.47 vs. limit=2.0 2022-12-07 04:40:49,649 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=14.11 vs. limit=2.0 2022-12-07 04:41:00,140 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=17.38 vs. limit=5.0 2022-12-07 04:41:12,192 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:41:13,361 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=30.62 vs. limit=5.0 2022-12-07 04:41:33,207 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:41:43,792 INFO [train.py:873] (1/4) Epoch 1, batch 400, loss[loss=0.4693, simple_loss=0.3911, pruned_loss=0.4184, over 14206.00 frames. ], tot_loss[loss=0.5313, simple_loss=0.4567, pruned_loss=0.4959, over 1712117.13 frames. ], batch size: 60, lr: 4.50e-02, grad_scale: 1.0 2022-12-07 04:41:47,368 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 2.807e+01 5.021e+01 6.661e+01 9.478e+01 2.713e+02, threshold=1.332e+02, percent-clipped=6.0 2022-12-07 04:41:48,627 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=14.59 vs. limit=5.0 2022-12-07 04:41:49,999 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=17.57 vs. limit=5.0 2022-12-07 04:42:10,509 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:42:14,011 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=16.72 vs. limit=5.0 2022-12-07 04:42:17,130 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:42:37,408 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.69 vs. limit=2.0 2022-12-07 04:42:55,425 INFO [train.py:873] (1/4) Epoch 1, batch 500, loss[loss=0.4515, simple_loss=0.3767, pruned_loss=0.3749, over 6928.00 frames. ], tot_loss[loss=0.503, simple_loss=0.4267, pruned_loss=0.4557, over 1795305.56 frames. ], batch size: 100, lr: 4.99e-02, grad_scale: 1.0 2022-12-07 04:42:59,042 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 2.328e+01 4.979e+01 6.873e+01 9.259e+01 1.851e+02, threshold=1.375e+02, percent-clipped=7.0 2022-12-07 04:43:20,960 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.82 vs. limit=5.0 2022-12-07 04:43:38,451 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:43:42,832 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:43:54,972 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8580, 3.2386, 3.3970, 3.6670, 3.3131, 3.3403, 2.6440, 3.6250], device='cuda:1'), covar=tensor([0.2129, 0.1346, 0.0766, 0.0735, 0.1317, 0.1198, 0.1314, 0.0773], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0014, 0.0014, 0.0013, 0.0014, 0.0014, 0.0014, 0.0013], device='cuda:1'), out_proj_covar=tensor([1.3605e-05, 1.3034e-05, 1.3231e-05, 1.2763e-05, 1.2371e-05, 1.3219e-05, 1.2961e-05, 1.2501e-05], device='cuda:1') 2022-12-07 04:43:57,705 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:44:05,266 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:44:05,749 INFO [train.py:873] (1/4) Epoch 1, batch 600, loss[loss=0.3693, simple_loss=0.3254, pruned_loss=0.2504, over 1335.00 frames. ], tot_loss[loss=0.4853, simple_loss=0.4065, pruned_loss=0.4265, over 1855011.06 frames. ], batch size: 100, lr: 4.98e-02, grad_scale: 1.0 2022-12-07 04:44:09,407 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 4.092e+01 6.985e+01 9.438e+01 1.366e+02 4.481e+02, threshold=1.888e+02, percent-clipped=23.0 2022-12-07 04:44:10,179 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=15.25 vs. limit=5.0 2022-12-07 04:44:13,634 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.74 vs. limit=5.0 2022-12-07 04:44:21,345 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:44:22,106 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.11 vs. limit=2.0 2022-12-07 04:44:25,262 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:44:39,013 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:44:41,261 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:44:41,811 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:45:15,578 INFO [train.py:873] (1/4) Epoch 1, batch 700, loss[loss=0.445, simple_loss=0.3611, pruned_loss=0.3544, over 11905.00 frames. ], tot_loss[loss=0.4742, simple_loss=0.3929, pruned_loss=0.4045, over 1850347.18 frames. ], batch size: 100, lr: 4.98e-02, grad_scale: 1.0 2022-12-07 04:45:19,021 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 4.942e+01 9.279e+01 1.429e+02 2.357e+02 5.770e+02, threshold=2.857e+02, percent-clipped=36.0 2022-12-07 04:45:38,367 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.34 vs. limit=2.0 2022-12-07 04:45:42,037 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:45:45,170 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:46:06,567 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.15 vs. limit=5.0 2022-12-07 04:46:15,740 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:46:25,440 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3299, 3.1902, 3.0152, 3.2722, 3.3592, 3.2404, 3.1813, 2.8913], device='cuda:1'), covar=tensor([0.1424, 0.2032, 0.1945, 0.1173, 0.1241, 0.1701, 0.1387, 0.1755], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0017, 0.0015, 0.0014, 0.0015, 0.0016, 0.0015, 0.0015], device='cuda:1'), out_proj_covar=tensor([1.1992e-05, 1.4766e-05, 1.2226e-05, 1.0865e-05, 1.2227e-05, 1.3683e-05, 1.1399e-05, 1.3274e-05], device='cuda:1') 2022-12-07 04:46:25,999 INFO [train.py:873] (1/4) Epoch 1, batch 800, loss[loss=0.4439, simple_loss=0.3633, pruned_loss=0.3331, over 14153.00 frames. ], tot_loss[loss=0.4588, simple_loss=0.3776, pruned_loss=0.3769, over 1899895.32 frames. ], batch size: 99, lr: 4.97e-02, grad_scale: 2.0 2022-12-07 04:46:29,401 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 4.458e+01 1.291e+02 1.818e+02 2.513e+02 6.152e+02, threshold=3.636e+02, percent-clipped=18.0 2022-12-07 04:46:31,529 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.18 vs. limit=5.0 2022-12-07 04:46:58,651 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:47:10,741 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.70 vs. limit=2.0 2022-12-07 04:47:36,964 INFO [train.py:873] (1/4) Epoch 1, batch 900, loss[loss=0.3818, simple_loss=0.3158, pruned_loss=0.2713, over 4919.00 frames. ], tot_loss[loss=0.4457, simple_loss=0.3659, pruned_loss=0.3517, over 1920471.10 frames. ], batch size: 100, lr: 4.96e-02, grad_scale: 2.0 2022-12-07 04:47:40,262 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.004e+01 1.837e+02 2.554e+02 3.837e+02 9.711e+02, threshold=5.109e+02, percent-clipped=29.0 2022-12-07 04:47:41,878 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:47:48,589 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:47:52,686 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:48:08,931 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:48:12,987 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:48:46,726 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:48:47,390 INFO [train.py:873] (1/4) Epoch 1, batch 1000, loss[loss=0.4243, simple_loss=0.3428, pruned_loss=0.3038, over 13554.00 frames. ], tot_loss[loss=0.4303, simple_loss=0.3538, pruned_loss=0.325, over 1999253.99 frames. ], batch size: 100, lr: 4.95e-02, grad_scale: 2.0 2022-12-07 04:48:50,981 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.449e+02 3.112e+02 4.355e+02 1.197e+03, threshold=6.224e+02, percent-clipped=18.0 2022-12-07 04:49:17,151 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:49:51,009 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:49:58,567 INFO [train.py:873] (1/4) Epoch 1, batch 1100, loss[loss=0.3832, simple_loss=0.3207, pruned_loss=0.2532, over 6945.00 frames. ], tot_loss[loss=0.4151, simple_loss=0.3419, pruned_loss=0.3022, over 1990463.61 frames. ], batch size: 100, lr: 4.94e-02, grad_scale: 2.0 2022-12-07 04:50:02,057 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.129e+02 2.767e+02 3.669e+02 5.185e+02 1.491e+03, threshold=7.337e+02, percent-clipped=11.0 2022-12-07 04:50:35,527 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:51:09,752 INFO [train.py:873] (1/4) Epoch 1, batch 1200, loss[loss=0.2876, simple_loss=0.2326, pruned_loss=0.1949, over 2627.00 frames. ], tot_loss[loss=0.4041, simple_loss=0.3339, pruned_loss=0.2838, over 1986503.64 frames. ], batch size: 100, lr: 4.93e-02, grad_scale: 4.0 2022-12-07 04:51:11,130 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:51:12,882 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.090e+02 2.727e+02 3.697e+02 4.642e+02 1.114e+03, threshold=7.394e+02, percent-clipped=5.0 2022-12-07 04:51:18,461 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:51:18,652 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.95 vs. limit=5.0 2022-12-07 04:51:21,890 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:51:25,879 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:51:41,230 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:51:55,334 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:51:59,302 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:52:03,602 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 04:52:12,040 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 04:52:15,061 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:52:20,508 INFO [train.py:873] (1/4) Epoch 1, batch 1300, loss[loss=0.4018, simple_loss=0.3305, pruned_loss=0.2608, over 14173.00 frames. ], tot_loss[loss=0.3923, simple_loss=0.3254, pruned_loss=0.2665, over 2031758.64 frames. ], batch size: 84, lr: 4.92e-02, grad_scale: 4.0 2022-12-07 04:52:23,970 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.849e+02 3.755e+02 4.812e+02 1.034e+03, threshold=7.510e+02, percent-clipped=8.0 2022-12-07 04:53:33,568 INFO [train.py:873] (1/4) Epoch 1, batch 1400, loss[loss=0.3969, simple_loss=0.3273, pruned_loss=0.2522, over 14270.00 frames. ], tot_loss[loss=0.3858, simple_loss=0.3206, pruned_loss=0.2551, over 2043703.67 frames. ], batch size: 76, lr: 4.91e-02, grad_scale: 4.0 2022-12-07 04:53:36,992 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 3.437e+02 4.813e+02 6.365e+02 1.224e+03, threshold=9.626e+02, percent-clipped=13.0 2022-12-07 04:53:38,899 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.72 vs. limit=5.0 2022-12-07 04:53:51,748 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.95 vs. limit=5.0 2022-12-07 04:53:57,795 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:54:07,291 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1514, 1.7771, 2.1967, 2.1822, 2.1448, 2.0683, 1.5206, 2.0251], device='cuda:1'), covar=tensor([0.0467, 0.0959, 0.0701, 0.0591, 0.0641, 0.0585, 0.1991, 0.0809], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0030, 0.0030, 0.0028, 0.0029, 0.0028, 0.0037, 0.0034], device='cuda:1'), out_proj_covar=tensor([2.1493e-05, 2.3547e-05, 2.3756e-05, 2.2414e-05, 2.5210e-05, 2.1701e-05, 3.8109e-05, 2.7000e-05], device='cuda:1') 2022-12-07 04:54:41,792 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:54:44,861 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2022-12-07 04:54:46,085 INFO [train.py:873] (1/4) Epoch 1, batch 1500, loss[loss=0.392, simple_loss=0.3299, pruned_loss=0.2397, over 14663.00 frames. ], tot_loss[loss=0.3765, simple_loss=0.3139, pruned_loss=0.2429, over 2030504.09 frames. ], batch size: 33, lr: 4.89e-02, grad_scale: 4.0 2022-12-07 04:54:47,694 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:54:49,749 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.091e+01 2.709e+02 3.903e+02 5.151e+02 1.116e+03, threshold=7.805e+02, percent-clipped=3.0 2022-12-07 04:54:52,100 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:55:22,614 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:55:29,970 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6136, 4.6563, 4.7898, 3.2404, 3.6262, 2.5674, 4.8360, 4.4061], device='cuda:1'), covar=tensor([0.0264, 0.0202, 0.0208, 0.1931, 0.0421, 0.1464, 0.0186, 0.0227], device='cuda:1'), in_proj_covar=tensor([0.0028, 0.0024, 0.0026, 0.0035, 0.0023, 0.0037, 0.0029, 0.0023], device='cuda:1'), out_proj_covar=tensor([2.0663e-05, 1.7845e-05, 1.9261e-05, 3.3761e-05, 1.8062e-05, 3.2479e-05, 2.3231e-05, 1.6309e-05], device='cuda:1') 2022-12-07 04:55:33,694 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:55:46,076 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 04:55:59,957 INFO [train.py:873] (1/4) Epoch 1, batch 1600, loss[loss=0.4058, simple_loss=0.3318, pruned_loss=0.252, over 13925.00 frames. ], tot_loss[loss=0.3706, simple_loss=0.3095, pruned_loss=0.234, over 2052204.57 frames. ], batch size: 23, lr: 4.88e-02, grad_scale: 8.0 2022-12-07 04:56:03,633 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 3.341e+02 4.121e+02 5.397e+02 1.445e+03, threshold=8.242e+02, percent-clipped=8.0 2022-12-07 04:56:07,802 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:56:11,472 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.12 vs. limit=2.0 2022-12-07 04:56:19,593 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:56:26,703 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.51 vs. limit=5.0 2022-12-07 04:56:52,734 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:56:54,655 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.07 vs. limit=5.0 2022-12-07 04:56:57,836 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.47 vs. limit=5.0 2022-12-07 04:57:13,820 INFO [train.py:873] (1/4) Epoch 1, batch 1700, loss[loss=0.2671, simple_loss=0.2175, pruned_loss=0.1642, over 1236.00 frames. ], tot_loss[loss=0.3638, simple_loss=0.305, pruned_loss=0.2253, over 2002477.19 frames. ], batch size: 100, lr: 4.86e-02, grad_scale: 8.0 2022-12-07 04:57:17,701 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 3.504e+02 4.735e+02 5.720e+02 1.294e+03, threshold=9.470e+02, percent-clipped=11.0 2022-12-07 04:58:21,543 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 04:58:29,845 INFO [train.py:873] (1/4) Epoch 1, batch 1800, loss[loss=0.3592, simple_loss=0.305, pruned_loss=0.2104, over 14242.00 frames. ], tot_loss[loss=0.3595, simple_loss=0.3023, pruned_loss=0.2187, over 2009972.61 frames. ], batch size: 80, lr: 4.85e-02, grad_scale: 8.0 2022-12-07 04:58:33,678 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.780e+01 3.438e+02 4.789e+02 6.464e+02 1.407e+03, threshold=9.578e+02, percent-clipped=4.0 2022-12-07 04:58:36,018 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:58:57,970 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:59:12,102 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:59:12,213 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:59:21,929 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:59:43,717 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:59:45,047 INFO [train.py:873] (1/4) Epoch 1, batch 1900, loss[loss=0.3072, simple_loss=0.26, pruned_loss=0.1787, over 4926.00 frames. ], tot_loss[loss=0.3551, simple_loss=0.2991, pruned_loss=0.2129, over 1992946.30 frames. ], batch size: 100, lr: 4.83e-02, grad_scale: 8.0 2022-12-07 04:59:48,578 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.566e+01 3.394e+02 4.356e+02 5.869e+02 1.320e+03, threshold=8.711e+02, percent-clipped=5.0 2022-12-07 04:59:58,571 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 05:00:01,429 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:00:08,244 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 05:00:16,900 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 05:00:35,448 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:00:55,558 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3357, 3.0288, 3.3969, 3.1691, 3.3635, 3.4594, 3.5904, 3.5622], device='cuda:1'), covar=tensor([0.0454, 0.0647, 0.0438, 0.0533, 0.0521, 0.0470, 0.0360, 0.0450], device='cuda:1'), in_proj_covar=tensor([0.0051, 0.0051, 0.0048, 0.0049, 0.0054, 0.0048, 0.0045, 0.0045], device='cuda:1'), out_proj_covar=tensor([5.0645e-05, 4.3262e-05, 4.5889e-05, 4.3947e-05, 4.7318e-05, 4.2940e-05, 3.9347e-05, 4.2550e-05], device='cuda:1') 2022-12-07 05:01:00,053 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2423, 3.8554, 4.3440, 4.4326, 4.3323, 4.4856, 4.5368, 4.5227], device='cuda:1'), covar=tensor([0.0456, 0.0660, 0.0411, 0.0235, 0.0439, 0.0429, 0.0329, 0.0561], device='cuda:1'), in_proj_covar=tensor([0.0050, 0.0050, 0.0048, 0.0048, 0.0053, 0.0047, 0.0044, 0.0044], device='cuda:1'), out_proj_covar=tensor([5.0071e-05, 4.2803e-05, 4.5032e-05, 4.3746e-05, 4.6227e-05, 4.2189e-05, 3.8589e-05, 4.1796e-05], device='cuda:1') 2022-12-07 05:01:01,972 INFO [train.py:873] (1/4) Epoch 1, batch 2000, loss[loss=0.3589, simple_loss=0.279, pruned_loss=0.2194, over 3866.00 frames. ], tot_loss[loss=0.3511, simple_loss=0.2962, pruned_loss=0.2079, over 1952471.77 frames. ], batch size: 100, lr: 4.82e-02, grad_scale: 8.0 2022-12-07 05:01:05,829 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.042e+02 3.554e+02 4.725e+02 6.724e+02 1.490e+03, threshold=9.451e+02, percent-clipped=5.0 2022-12-07 05:01:49,197 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.71 vs. limit=2.0 2022-12-07 05:02:12,117 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2022-12-07 05:02:14,870 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 05:02:23,810 INFO [train.py:873] (1/4) Epoch 1, batch 2100, loss[loss=0.3358, simple_loss=0.2929, pruned_loss=0.1893, over 14528.00 frames. ], tot_loss[loss=0.3426, simple_loss=0.2923, pruned_loss=0.1992, over 2027614.60 frames. ], batch size: 49, lr: 4.80e-02, grad_scale: 16.0 2022-12-07 05:02:27,744 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 4.809e+01 3.029e+02 3.794e+02 4.821e+02 1.470e+03, threshold=7.588e+02, percent-clipped=1.0 2022-12-07 05:02:28,939 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2022-12-07 05:02:53,154 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:03:31,519 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.55 vs. limit=2.0 2022-12-07 05:03:38,928 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:03:45,041 INFO [train.py:873] (1/4) Epoch 1, batch 2200, loss[loss=0.2911, simple_loss=0.2386, pruned_loss=0.1718, over 3898.00 frames. ], tot_loss[loss=0.34, simple_loss=0.2907, pruned_loss=0.1964, over 1983279.46 frames. ], batch size: 100, lr: 4.78e-02, grad_scale: 16.0 2022-12-07 05:03:49,158 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 3.028e+02 4.377e+02 7.241e+02 2.014e+03, threshold=8.755e+02, percent-clipped=23.0 2022-12-07 05:03:54,587 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:00,673 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4251, 3.1052, 3.0837, 3.0787, 3.2757, 3.0526, 3.5526, 3.3835], device='cuda:1'), covar=tensor([0.0607, 0.0810, 0.0758, 0.0628, 0.0546, 0.0590, 0.0437, 0.0635], device='cuda:1'), in_proj_covar=tensor([0.0043, 0.0037, 0.0047, 0.0040, 0.0047, 0.0038, 0.0037, 0.0044], device='cuda:1'), out_proj_covar=tensor([4.5239e-05, 4.1009e-05, 4.9415e-05, 4.0996e-05, 4.8873e-05, 3.8720e-05, 4.2606e-05, 4.4238e-05], device='cuda:1') 2022-12-07 05:04:01,874 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 05:04:05,010 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:37,950 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8994, 2.0078, 2.9086, 2.3663, 2.9882, 2.9620, 2.2324, 1.9770], device='cuda:1'), covar=tensor([0.0213, 0.1591, 0.0174, 0.0428, 0.0281, 0.0211, 0.0530, 0.1915], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0047, 0.0023, 0.0031, 0.0026, 0.0026, 0.0022, 0.0051], device='cuda:1'), out_proj_covar=tensor([1.7360e-05, 4.5069e-05, 1.4803e-05, 2.2437e-05, 1.6339e-05, 1.6738e-05, 1.6007e-05, 4.4693e-05], device='cuda:1') 2022-12-07 05:04:38,703 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:04:41,028 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:58,901 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.22 vs. limit=5.0 2022-12-07 05:05:06,465 INFO [train.py:873] (1/4) Epoch 1, batch 2300, loss[loss=0.3189, simple_loss=0.2729, pruned_loss=0.1825, over 14217.00 frames. ], tot_loss[loss=0.3323, simple_loss=0.2863, pruned_loss=0.1902, over 1968729.60 frames. ], batch size: 35, lr: 4.77e-02, grad_scale: 16.0 2022-12-07 05:05:10,739 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.291e+01 3.360e+02 4.447e+02 6.275e+02 1.931e+03, threshold=8.894e+02, percent-clipped=10.0 2022-12-07 05:05:15,456 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 05:05:18,536 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:05:38,623 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:05:48,281 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2215, 3.7043, 4.9667, 4.3949, 4.9055, 4.6000, 4.1618, 4.5277], device='cuda:1'), covar=tensor([0.0239, 0.0610, 0.0183, 0.0295, 0.0162, 0.0135, 0.0357, 0.0293], device='cuda:1'), in_proj_covar=tensor([0.0034, 0.0037, 0.0035, 0.0032, 0.0036, 0.0028, 0.0036, 0.0039], device='cuda:1'), out_proj_covar=tensor([3.2038e-05, 3.5871e-05, 3.4543e-05, 3.0578e-05, 3.5220e-05, 2.7673e-05, 3.4050e-05, 3.9657e-05], device='cuda:1') 2022-12-07 05:05:52,013 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 05:05:55,007 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.72 vs. limit=2.0 2022-12-07 05:06:28,678 INFO [train.py:873] (1/4) Epoch 1, batch 2400, loss[loss=0.3112, simple_loss=0.2688, pruned_loss=0.1768, over 6919.00 frames. ], tot_loss[loss=0.3281, simple_loss=0.2842, pruned_loss=0.1866, over 1979984.23 frames. ], batch size: 100, lr: 4.75e-02, grad_scale: 16.0 2022-12-07 05:06:28,811 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:06:32,586 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 3.236e+02 4.365e+02 6.045e+02 1.327e+03, threshold=8.731e+02, percent-clipped=8.0 2022-12-07 05:06:59,993 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1342, 1.4574, 1.8628, 1.8804, 1.8232, 1.6599, 1.3632, 1.8909], device='cuda:1'), covar=tensor([0.0931, 0.0609, 0.0199, 0.0196, 0.0249, 0.0279, 0.0890, 0.0226], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0026, 0.0020, 0.0021, 0.0020, 0.0021, 0.0026, 0.0024], device='cuda:1'), out_proj_covar=tensor([2.5468e-05, 2.2859e-05, 1.5896e-05, 1.5707e-05, 1.5478e-05, 1.6917e-05, 2.3418e-05, 1.9480e-05], device='cuda:1') 2022-12-07 05:07:32,702 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 05:07:43,681 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-07 05:07:44,844 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:07:50,098 INFO [train.py:873] (1/4) Epoch 1, batch 2500, loss[loss=0.2806, simple_loss=0.244, pruned_loss=0.1585, over 3850.00 frames. ], tot_loss[loss=0.3246, simple_loss=0.2824, pruned_loss=0.1838, over 1940282.37 frames. ], batch size: 100, lr: 4.73e-02, grad_scale: 16.0 2022-12-07 05:07:52,137 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.62 vs. limit=2.0 2022-12-07 05:07:54,361 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.013e+02 3.632e+02 4.712e+02 6.167e+02 1.692e+03, threshold=9.425e+02, percent-clipped=8.0 2022-12-07 05:07:54,512 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6246, 4.9008, 5.5066, 5.6206, 5.5044, 5.0001, 5.0651, 5.2025], device='cuda:1'), covar=tensor([0.0158, 0.0293, 0.0189, 0.0350, 0.0190, 0.0254, 0.0329, 0.0358], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0044, 0.0035, 0.0032, 0.0035, 0.0036, 0.0040, 0.0040], device='cuda:1'), out_proj_covar=tensor([3.2723e-05, 4.5324e-05, 3.2784e-05, 3.3219e-05, 3.4358e-05, 3.9178e-05, 4.0043e-05, 4.2170e-05], device='cuda:1') 2022-12-07 05:08:00,197 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:11,272 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:23,747 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:25,501 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.56 vs. limit=5.0 2022-12-07 05:08:27,034 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1338, 4.9157, 4.8467, 4.7567, 5.1720, 4.2463, 5.4348, 5.0975], device='cuda:1'), covar=tensor([0.0582, 0.0500, 0.0627, 0.0597, 0.0418, 0.0695, 0.0381, 0.0493], device='cuda:1'), in_proj_covar=tensor([0.0048, 0.0039, 0.0055, 0.0046, 0.0053, 0.0042, 0.0043, 0.0052], device='cuda:1'), out_proj_covar=tensor([5.3988e-05, 4.4146e-05, 6.0111e-05, 5.0945e-05, 5.7104e-05, 4.4675e-05, 5.1703e-05, 5.5891e-05], device='cuda:1') 2022-12-07 05:08:39,730 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:49,999 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:58,405 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.49 vs. limit=2.0 2022-12-07 05:09:13,572 INFO [train.py:873] (1/4) Epoch 1, batch 2600, loss[loss=0.3149, simple_loss=0.2844, pruned_loss=0.1727, over 13927.00 frames. ], tot_loss[loss=0.3208, simple_loss=0.28, pruned_loss=0.181, over 1953752.72 frames. ], batch size: 23, lr: 4.71e-02, grad_scale: 16.0 2022-12-07 05:09:17,459 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.018e+02 2.998e+02 4.312e+02 6.612e+02 1.607e+03, threshold=8.624e+02, percent-clipped=4.0 2022-12-07 05:10:11,441 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.06 vs. limit=5.0 2022-12-07 05:10:18,320 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2022-12-07 05:10:20,885 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.14 vs. limit=2.0 2022-12-07 05:10:23,665 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:10:32,516 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:10:36,428 INFO [train.py:873] (1/4) Epoch 1, batch 2700, loss[loss=0.374, simple_loss=0.3119, pruned_loss=0.2181, over 14292.00 frames. ], tot_loss[loss=0.3176, simple_loss=0.2785, pruned_loss=0.1785, over 1995589.47 frames. ], batch size: 35, lr: 4.69e-02, grad_scale: 16.0 2022-12-07 05:10:40,371 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 3.041e+02 4.444e+02 6.160e+02 1.367e+03, threshold=8.889e+02, percent-clipped=4.0 2022-12-07 05:11:14,276 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 05:11:18,164 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.30 vs. limit=5.0 2022-12-07 05:11:38,737 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 05:11:50,720 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7768, 4.6159, 2.7868, 5.0259, 4.3811, 4.8451, 4.4163, 2.2165], device='cuda:1'), covar=tensor([0.0103, 0.0218, 0.2442, 0.0096, 0.0124, 0.0095, 0.0173, 0.2468], device='cuda:1'), in_proj_covar=tensor([0.0048, 0.0052, 0.0085, 0.0044, 0.0040, 0.0048, 0.0055, 0.0106], device='cuda:1'), out_proj_covar=tensor([2.3709e-05, 2.8200e-05, 6.4610e-05, 2.3102e-05, 2.0665e-05, 2.4107e-05, 2.8306e-05, 8.2259e-05], device='cuda:1') 2022-12-07 05:12:00,035 INFO [train.py:873] (1/4) Epoch 1, batch 2800, loss[loss=0.313, simple_loss=0.2762, pruned_loss=0.1749, over 14241.00 frames. ], tot_loss[loss=0.317, simple_loss=0.2786, pruned_loss=0.1778, over 2043954.79 frames. ], batch size: 69, lr: 4.67e-02, grad_scale: 8.0 2022-12-07 05:12:05,006 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 3.640e+02 4.591e+02 6.544e+02 1.997e+03, threshold=9.181e+02, percent-clipped=12.0 2022-12-07 05:12:07,254 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.80 vs. limit=2.0 2022-12-07 05:12:18,756 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:12:32,865 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 05:12:34,815 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2494, 1.8294, 2.5013, 2.1984, 1.9989, 2.2219, 1.6860, 2.1574], device='cuda:1'), covar=tensor([0.0206, 0.0789, 0.0145, 0.0336, 0.0243, 0.0243, 0.0684, 0.0280], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0042, 0.0027, 0.0030, 0.0034, 0.0029, 0.0028, 0.0028], device='cuda:1'), out_proj_covar=tensor([2.8471e-05, 4.3294e-05, 2.8565e-05, 2.8533e-05, 3.2947e-05, 2.6871e-05, 2.9178e-05, 2.7862e-05], device='cuda:1') 2022-12-07 05:12:55,376 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 05:13:09,567 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:13:12,867 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1729, 1.5832, 1.7600, 1.9215, 2.0476, 1.7943, 2.0374, 1.5825], device='cuda:1'), covar=tensor([0.0149, 0.0269, 0.0216, 0.0202, 0.0197, 0.0220, 0.0165, 0.0199], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0019, 0.0021, 0.0017, 0.0019, 0.0021, 0.0017, 0.0019], device='cuda:1'), out_proj_covar=tensor([1.7004e-05, 1.6407e-05, 1.7356e-05, 1.4583e-05, 1.6563e-05, 1.7338e-05, 1.5056e-05, 1.6088e-05], device='cuda:1') 2022-12-07 05:13:22,398 INFO [train.py:873] (1/4) Epoch 1, batch 2900, loss[loss=0.2954, simple_loss=0.2598, pruned_loss=0.1655, over 5987.00 frames. ], tot_loss[loss=0.3153, simple_loss=0.2772, pruned_loss=0.1767, over 1989031.19 frames. ], batch size: 100, lr: 4.65e-02, grad_scale: 8.0 2022-12-07 05:13:27,203 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 3.059e+02 4.261e+02 6.986e+02 2.302e+03, threshold=8.522e+02, percent-clipped=12.0 2022-12-07 05:13:49,130 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 05:13:52,864 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:14:26,293 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5992, 3.8035, 3.0046, 3.2000, 3.6257, 2.4002, 3.2006, 2.2513], device='cuda:1'), covar=tensor([0.0394, 0.0293, 0.0534, 0.0562, 0.0290, 0.2402, 0.0458, 0.2185], device='cuda:1'), in_proj_covar=tensor([0.0029, 0.0029, 0.0023, 0.0028, 0.0032, 0.0048, 0.0025, 0.0045], device='cuda:1'), out_proj_covar=tensor([2.0006e-05, 1.9440e-05, 1.4973e-05, 1.9418e-05, 2.1799e-05, 4.1782e-05, 1.8256e-05, 3.6883e-05], device='cuda:1') 2022-12-07 05:14:42,998 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:14:45,324 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 05:14:48,186 INFO [train.py:873] (1/4) Epoch 1, batch 3000, loss[loss=0.295, simple_loss=0.2679, pruned_loss=0.161, over 13957.00 frames. ], tot_loss[loss=0.3112, simple_loss=0.2744, pruned_loss=0.1741, over 1905125.37 frames. ], batch size: 26, lr: 4.63e-02, grad_scale: 8.0 2022-12-07 05:14:48,186 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 05:14:56,535 INFO [train.py:905] (1/4) Epoch 1, validation: loss=0.2054, simple_loss=0.2366, pruned_loss=0.08706, over 857387.00 frames. 2022-12-07 05:14:56,536 INFO [train.py:906] (1/4) Maximum memory allocated so far is 16812MB 2022-12-07 05:15:01,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.332e+02 3.340e+02 4.389e+02 5.804e+02 1.068e+03, threshold=8.777e+02, percent-clipped=8.0 2022-12-07 05:15:05,935 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 05:15:11,897 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:15:30,354 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:15:32,924 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:16:04,050 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:16:08,887 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9324, 1.9051, 2.5879, 1.7189, 1.9467, 2.0504, 1.4778, 2.4696], device='cuda:1'), covar=tensor([0.0278, 0.2695, 0.0427, 0.1262, 0.0408, 0.0949, 0.1739, 0.0675], device='cuda:1'), in_proj_covar=tensor([0.0028, 0.0049, 0.0033, 0.0041, 0.0028, 0.0039, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([2.2065e-05, 4.5359e-05, 2.6310e-05, 3.7708e-05, 2.3150e-05, 3.6865e-05, 2.7136e-05, 2.7464e-05], device='cuda:1') 2022-12-07 05:16:21,381 INFO [train.py:873] (1/4) Epoch 1, batch 3100, loss[loss=0.3462, simple_loss=0.2927, pruned_loss=0.1999, over 8650.00 frames. ], tot_loss[loss=0.3089, simple_loss=0.2734, pruned_loss=0.1722, over 1946969.37 frames. ], batch size: 100, lr: 4.61e-02, grad_scale: 8.0 2022-12-07 05:16:26,603 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 3.492e+02 4.643e+02 6.451e+02 1.266e+03, threshold=9.287e+02, percent-clipped=10.0 2022-12-07 05:17:04,498 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2022-12-07 05:17:07,530 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3879, 3.6035, 2.1700, 3.5405, 3.1141, 3.8065, 3.2160, 2.0698], device='cuda:1'), covar=tensor([0.0262, 0.0322, 0.3301, 0.0131, 0.0194, 0.0192, 0.0308, 0.2838], device='cuda:1'), in_proj_covar=tensor([0.0058, 0.0064, 0.0112, 0.0050, 0.0046, 0.0054, 0.0064, 0.0126], device='cuda:1'), out_proj_covar=tensor([2.8252e-05, 3.4528e-05, 7.8433e-05, 2.6261e-05, 2.3669e-05, 2.7131e-05, 3.3547e-05, 9.1590e-05], device='cuda:1') 2022-12-07 05:17:21,268 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.86 vs. limit=2.0 2022-12-07 05:17:27,823 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:17:45,613 INFO [train.py:873] (1/4) Epoch 1, batch 3200, loss[loss=0.2943, simple_loss=0.2579, pruned_loss=0.1653, over 6969.00 frames. ], tot_loss[loss=0.309, simple_loss=0.2734, pruned_loss=0.1723, over 1885609.98 frames. ], batch size: 100, lr: 4.59e-02, grad_scale: 8.0 2022-12-07 05:17:50,388 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 3.868e+02 5.118e+02 7.409e+02 2.192e+03, threshold=1.024e+03, percent-clipped=11.0 2022-12-07 05:18:57,691 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:19:00,899 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7592, 1.5986, 3.6326, 3.5354, 3.7195, 3.5226, 2.2549, 3.7248], device='cuda:1'), covar=tensor([0.0186, 0.2110, 0.0274, 0.0344, 0.0203, 0.0226, 0.0743, 0.0248], device='cuda:1'), in_proj_covar=tensor([0.0032, 0.0052, 0.0030, 0.0037, 0.0038, 0.0032, 0.0032, 0.0030], device='cuda:1'), out_proj_covar=tensor([3.1648e-05, 5.7935e-05, 3.4536e-05, 3.9060e-05, 3.8944e-05, 3.3706e-05, 3.6803e-05, 3.2887e-05], device='cuda:1') 2022-12-07 05:19:01,700 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:19:06,306 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.14 vs. limit=2.0 2022-12-07 05:19:08,886 INFO [train.py:873] (1/4) Epoch 1, batch 3300, loss[loss=0.284, simple_loss=0.2262, pruned_loss=0.1709, over 1163.00 frames. ], tot_loss[loss=0.3051, simple_loss=0.2709, pruned_loss=0.1696, over 1852393.47 frames. ], batch size: 100, lr: 4.57e-02, grad_scale: 8.0 2022-12-07 05:19:14,103 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.921e+02 3.689e+02 4.978e+02 9.002e+02, threshold=7.379e+02, percent-clipped=0.0 2022-12-07 05:19:42,806 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:19:49,883 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 05:20:11,313 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:20:18,626 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2022-12-07 05:20:22,021 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-07 05:20:23,310 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0082, 1.4952, 2.0995, 1.8446, 2.3621, 1.5929, 1.9505, 1.5203], device='cuda:1'), covar=tensor([0.0370, 0.0767, 0.0387, 0.0568, 0.0312, 0.0653, 0.0560, 0.0443], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0023, 0.0026, 0.0022, 0.0021, 0.0026, 0.0020, 0.0024], device='cuda:1'), out_proj_covar=tensor([2.0776e-05, 2.1287e-05, 2.3608e-05, 2.1225e-05, 1.9215e-05, 2.3290e-05, 1.7872e-05, 2.1693e-05], device='cuda:1') 2022-12-07 05:20:24,037 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:20:29,720 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 05:20:34,759 INFO [train.py:873] (1/4) Epoch 1, batch 3400, loss[loss=0.3019, simple_loss=0.2803, pruned_loss=0.1618, over 14673.00 frames. ], tot_loss[loss=0.3033, simple_loss=0.2697, pruned_loss=0.1685, over 1854164.52 frames. ], batch size: 23, lr: 4.55e-02, grad_scale: 8.0 2022-12-07 05:20:39,711 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.829e+02 3.330e+02 4.669e+02 8.149e+02 2.778e+03, threshold=9.337e+02, percent-clipped=27.0 2022-12-07 05:20:48,223 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.23 vs. limit=2.0 2022-12-07 05:21:43,545 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:22:01,456 INFO [train.py:873] (1/4) Epoch 1, batch 3500, loss[loss=0.2442, simple_loss=0.2092, pruned_loss=0.1397, over 2659.00 frames. ], tot_loss[loss=0.3006, simple_loss=0.2682, pruned_loss=0.1665, over 1865948.84 frames. ], batch size: 100, lr: 4.53e-02, grad_scale: 8.0 2022-12-07 05:22:06,643 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.311e+02 3.200e+02 4.145e+02 5.923e+02 9.666e+02, threshold=8.289e+02, percent-clipped=1.0 2022-12-07 05:22:22,294 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=25.40 vs. limit=5.0 2022-12-07 05:22:24,411 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:22:40,271 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:23:10,032 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7442, 5.1379, 5.3586, 5.9674, 5.9131, 5.5353, 5.4504, 5.5507], device='cuda:1'), covar=tensor([0.0208, 0.0305, 0.0227, 0.0211, 0.0153, 0.0262, 0.0231, 0.0209], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0056, 0.0044, 0.0042, 0.0041, 0.0045, 0.0054, 0.0051], device='cuda:1'), out_proj_covar=tensor([4.6884e-05, 6.7690e-05, 5.0186e-05, 5.2532e-05, 4.8267e-05, 5.7652e-05, 6.9382e-05, 6.3826e-05], device='cuda:1') 2022-12-07 05:23:21,051 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:23:28,042 INFO [train.py:873] (1/4) Epoch 1, batch 3600, loss[loss=0.2279, simple_loss=0.1921, pruned_loss=0.1318, over 1282.00 frames. ], tot_loss[loss=0.296, simple_loss=0.2653, pruned_loss=0.1633, over 1866452.05 frames. ], batch size: 100, lr: 4.50e-02, grad_scale: 8.0 2022-12-07 05:23:33,317 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 3.769e+02 4.638e+02 5.846e+02 1.150e+03, threshold=9.276e+02, percent-clipped=6.0 2022-12-07 05:23:33,563 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 05:24:02,790 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:24:05,388 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:24:22,460 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2022-12-07 05:24:29,591 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7743, 1.3454, 1.9186, 1.9167, 1.9452, 1.7759, 1.3966, 1.5184], device='cuda:1'), covar=tensor([0.0226, 0.0396, 0.0154, 0.0126, 0.0139, 0.0162, 0.0429, 0.0356], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0032, 0.0037, 0.0033, 0.0034, 0.0032, 0.0049, 0.0037], device='cuda:1'), out_proj_covar=tensor([3.0219e-05, 2.9608e-05, 3.2626e-05, 2.5492e-05, 2.8188e-05, 2.5572e-05, 5.1663e-05, 3.6111e-05], device='cuda:1') 2022-12-07 05:24:33,194 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:24:34,049 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:24:55,838 INFO [train.py:873] (1/4) Epoch 1, batch 3700, loss[loss=0.3098, simple_loss=0.2784, pruned_loss=0.1706, over 14085.00 frames. ], tot_loss[loss=0.2944, simple_loss=0.2641, pruned_loss=0.1624, over 1850155.93 frames. ], batch size: 26, lr: 4.48e-02, grad_scale: 8.0 2022-12-07 05:25:00,990 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 3.561e+02 4.722e+02 6.251e+02 1.502e+03, threshold=9.443e+02, percent-clipped=7.0 2022-12-07 05:25:14,828 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:25:26,796 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 05:25:58,646 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 05:26:07,184 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 05:26:22,263 INFO [train.py:873] (1/4) Epoch 1, batch 3800, loss[loss=0.2328, simple_loss=0.1986, pruned_loss=0.1334, over 2662.00 frames. ], tot_loss[loss=0.2939, simple_loss=0.2646, pruned_loss=0.1616, over 1890643.06 frames. ], batch size: 100, lr: 4.46e-02, grad_scale: 8.0 2022-12-07 05:26:27,325 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 3.207e+02 4.617e+02 6.095e+02 1.543e+03, threshold=9.233e+02, percent-clipped=5.0 2022-12-07 05:27:50,296 INFO [train.py:873] (1/4) Epoch 1, batch 3900, loss[loss=0.3085, simple_loss=0.2815, pruned_loss=0.1678, over 14305.00 frames. ], tot_loss[loss=0.2936, simple_loss=0.2642, pruned_loss=0.1615, over 1861161.78 frames. ], batch size: 37, lr: 4.44e-02, grad_scale: 8.0 2022-12-07 05:27:51,234 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:27:55,147 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 3.646e+02 5.236e+02 6.626e+02 1.873e+03, threshold=1.047e+03, percent-clipped=8.0 2022-12-07 05:28:11,778 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:28:16,584 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 05:28:27,559 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:28:52,103 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.13 vs. limit=2.0 2022-12-07 05:28:55,513 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.43 vs. limit=5.0 2022-12-07 05:28:56,869 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7994, 1.5905, 1.6977, 1.9713, 1.8711, 1.4339, 1.6393, 1.4955], device='cuda:1'), covar=tensor([0.0159, 0.0360, 0.0487, 0.0120, 0.0111, 0.0295, 0.0202, 0.0594], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0081, 0.0137, 0.0059, 0.0057, 0.0070, 0.0081, 0.0148], device='cuda:1'), out_proj_covar=tensor([3.5445e-05, 4.5418e-05, 8.7635e-05, 2.9769e-05, 2.9099e-05, 3.5262e-05, 4.2769e-05, 9.5997e-05], device='cuda:1') 2022-12-07 05:29:05,117 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:29:09,482 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:29:17,381 INFO [train.py:873] (1/4) Epoch 1, batch 4000, loss[loss=0.28, simple_loss=0.2384, pruned_loss=0.1609, over 3884.00 frames. ], tot_loss[loss=0.2909, simple_loss=0.2621, pruned_loss=0.1599, over 1772796.14 frames. ], batch size: 100, lr: 4.42e-02, grad_scale: 8.0 2022-12-07 05:29:22,734 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 3.697e+02 5.328e+02 7.328e+02 1.359e+03, threshold=1.066e+03, percent-clipped=6.0 2022-12-07 05:29:45,763 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:30:05,563 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.10 vs. limit=2.0 2022-12-07 05:30:12,074 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3435, 1.4550, 2.6412, 2.0628, 2.4217, 2.1415, 1.3985, 2.4627], device='cuda:1'), covar=tensor([0.0177, 0.1311, 0.0165, 0.0607, 0.0164, 0.0266, 0.0692, 0.0248], device='cuda:1'), in_proj_covar=tensor([0.0040, 0.0066, 0.0037, 0.0050, 0.0047, 0.0039, 0.0035, 0.0038], device='cuda:1'), out_proj_covar=tensor([4.7547e-05, 7.8917e-05, 4.6592e-05, 6.2762e-05, 5.6056e-05, 4.6217e-05, 4.6572e-05, 4.7079e-05], device='cuda:1') 2022-12-07 05:30:34,750 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2293, 3.5453, 3.6938, 4.0032, 4.2108, 4.1424, 3.5045, 3.0542], device='cuda:1'), covar=tensor([0.0254, 0.1685, 0.0179, 0.0310, 0.0178, 0.0213, 0.0229, 0.1753], device='cuda:1'), in_proj_covar=tensor([0.0057, 0.0099, 0.0044, 0.0052, 0.0046, 0.0053, 0.0043, 0.0100], device='cuda:1'), out_proj_covar=tensor([3.3489e-05, 7.5661e-05, 2.4637e-05, 3.2428e-05, 2.6511e-05, 3.1043e-05, 2.7646e-05, 7.3265e-05], device='cuda:1') 2022-12-07 05:30:47,315 INFO [train.py:873] (1/4) Epoch 1, batch 4100, loss[loss=0.3189, simple_loss=0.284, pruned_loss=0.1769, over 12737.00 frames. ], tot_loss[loss=0.288, simple_loss=0.2608, pruned_loss=0.1576, over 1878158.62 frames. ], batch size: 100, lr: 4.40e-02, grad_scale: 8.0 2022-12-07 05:30:52,473 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.129e+02 3.295e+02 4.605e+02 5.758e+02 1.082e+03, threshold=9.210e+02, percent-clipped=1.0 2022-12-07 05:31:22,868 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:31:32,274 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2022-12-07 05:32:16,951 INFO [train.py:873] (1/4) Epoch 1, batch 4200, loss[loss=0.2884, simple_loss=0.2538, pruned_loss=0.1615, over 9503.00 frames. ], tot_loss[loss=0.2862, simple_loss=0.26, pruned_loss=0.1561, over 1960014.54 frames. ], batch size: 100, lr: 4.38e-02, grad_scale: 8.0 2022-12-07 05:32:17,767 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:32:17,814 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:32:18,587 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7679, 2.6488, 2.4092, 2.6555, 2.4585, 2.2047, 2.7175, 2.7284], device='cuda:1'), covar=tensor([0.0764, 0.0714, 0.1070, 0.0726, 0.0957, 0.0782, 0.0886, 0.0960], device='cuda:1'), in_proj_covar=tensor([0.0067, 0.0054, 0.0076, 0.0069, 0.0074, 0.0050, 0.0070, 0.0074], device='cuda:1'), out_proj_covar=tensor([8.6379e-05, 7.0493e-05, 9.5905e-05, 8.9806e-05, 9.3200e-05, 6.3765e-05, 9.8810e-05, 9.8147e-05], device='cuda:1') 2022-12-07 05:32:21,867 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 3.430e+02 4.467e+02 5.944e+02 1.498e+03, threshold=8.934e+02, percent-clipped=5.0 2022-12-07 05:32:36,097 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.09 vs. limit=2.0 2022-12-07 05:32:38,937 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 05:33:00,185 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:33:23,334 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 05:33:28,876 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:33:34,072 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:33:45,949 INFO [train.py:873] (1/4) Epoch 1, batch 4300, loss[loss=0.272, simple_loss=0.2595, pruned_loss=0.1423, over 14270.00 frames. ], tot_loss[loss=0.2877, simple_loss=0.2612, pruned_loss=0.1571, over 1994118.28 frames. ], batch size: 57, lr: 4.35e-02, grad_scale: 8.0 2022-12-07 05:33:51,477 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.040e+01 3.141e+02 3.918e+02 5.765e+02 1.012e+03, threshold=7.835e+02, percent-clipped=6.0 2022-12-07 05:34:13,549 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:34:29,147 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:34:56,530 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:35:15,297 INFO [train.py:873] (1/4) Epoch 1, batch 4400, loss[loss=0.2643, simple_loss=0.2199, pruned_loss=0.1544, over 1268.00 frames. ], tot_loss[loss=0.2859, simple_loss=0.2594, pruned_loss=0.1562, over 1916919.17 frames. ], batch size: 100, lr: 4.33e-02, grad_scale: 8.0 2022-12-07 05:35:20,266 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.063e+02 3.285e+02 4.190e+02 5.766e+02 1.366e+03, threshold=8.380e+02, percent-clipped=8.0 2022-12-07 05:36:40,725 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:36:44,321 INFO [train.py:873] (1/4) Epoch 1, batch 4500, loss[loss=0.2839, simple_loss=0.2688, pruned_loss=0.1495, over 14003.00 frames. ], tot_loss[loss=0.2846, simple_loss=0.259, pruned_loss=0.1551, over 1955376.54 frames. ], batch size: 22, lr: 4.31e-02, grad_scale: 8.0 2022-12-07 05:36:49,368 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 3.529e+02 4.865e+02 6.809e+02 1.533e+03, threshold=9.730e+02, percent-clipped=12.0 2022-12-07 05:37:16,852 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3593, 2.6527, 2.7979, 2.9811, 2.5761, 2.2240, 2.6029, 1.9828], device='cuda:1'), covar=tensor([0.0393, 0.0302, 0.0137, 0.0247, 0.0229, 0.0987, 0.0170, 0.0900], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0041, 0.0027, 0.0034, 0.0040, 0.0062, 0.0030, 0.0055], device='cuda:1'), out_proj_covar=tensor([2.7680e-05, 2.9597e-05, 1.8711e-05, 2.4572e-05, 2.7694e-05, 5.0239e-05, 1.9361e-05, 4.4573e-05], device='cuda:1') 2022-12-07 05:37:29,416 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.39 vs. limit=2.0 2022-12-07 05:37:54,666 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:38:11,655 INFO [train.py:873] (1/4) Epoch 1, batch 4600, loss[loss=0.2708, simple_loss=0.2439, pruned_loss=0.1489, over 5953.00 frames. ], tot_loss[loss=0.2867, simple_loss=0.26, pruned_loss=0.1567, over 1941436.22 frames. ], batch size: 100, lr: 4.29e-02, grad_scale: 8.0 2022-12-07 05:38:16,961 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 3.319e+02 4.792e+02 5.771e+02 3.822e+03, threshold=9.585e+02, percent-clipped=7.0 2022-12-07 05:38:17,096 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0692, 2.8260, 3.0020, 2.7285, 2.9877, 2.9140, 3.0252, 2.9822], device='cuda:1'), covar=tensor([0.0275, 0.0505, 0.0346, 0.0370, 0.0358, 0.0430, 0.0606, 0.0494], device='cuda:1'), in_proj_covar=tensor([0.0077, 0.0072, 0.0085, 0.0058, 0.0082, 0.0077, 0.0079, 0.0067], device='cuda:1'), out_proj_covar=tensor([8.9511e-05, 8.5556e-05, 9.5573e-05, 6.7708e-05, 9.6039e-05, 8.6010e-05, 1.0141e-04, 8.2210e-05], device='cuda:1') 2022-12-07 05:38:37,717 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:38:49,789 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:39:41,708 INFO [train.py:873] (1/4) Epoch 1, batch 4700, loss[loss=0.2539, simple_loss=0.2464, pruned_loss=0.1307, over 14226.00 frames. ], tot_loss[loss=0.286, simple_loss=0.2594, pruned_loss=0.1563, over 1914695.97 frames. ], batch size: 32, lr: 4.27e-02, grad_scale: 8.0 2022-12-07 05:39:46,821 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 3.103e+02 4.134e+02 5.921e+02 2.901e+03, threshold=8.268e+02, percent-clipped=6.0 2022-12-07 05:40:07,929 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5669, 1.6168, 3.3897, 3.1724, 3.2042, 3.0002, 2.3993, 3.7274], device='cuda:1'), covar=tensor([0.1783, 0.1604, 0.0203, 0.0236, 0.0178, 0.0323, 0.0560, 0.0174], device='cuda:1'), in_proj_covar=tensor([0.0070, 0.0069, 0.0035, 0.0045, 0.0038, 0.0042, 0.0038, 0.0037], device='cuda:1'), out_proj_covar=tensor([9.2737e-05, 8.8929e-05, 4.5085e-05, 6.5054e-05, 4.5132e-05, 5.4054e-05, 5.1374e-05, 4.5041e-05], device='cuda:1') 2022-12-07 05:40:19,418 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.24 vs. limit=2.0 2022-12-07 05:41:06,515 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:41:10,381 INFO [train.py:873] (1/4) Epoch 1, batch 4800, loss[loss=0.2809, simple_loss=0.2426, pruned_loss=0.1596, over 3901.00 frames. ], tot_loss[loss=0.2841, simple_loss=0.2581, pruned_loss=0.155, over 1913948.56 frames. ], batch size: 100, lr: 4.25e-02, grad_scale: 16.0 2022-12-07 05:41:15,576 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 3.055e+02 4.514e+02 6.112e+02 1.452e+03, threshold=9.028e+02, percent-clipped=8.0 2022-12-07 05:41:47,181 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 05:41:47,675 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2363, 2.7397, 3.1712, 2.9255, 3.1940, 2.9751, 2.5701, 2.5557], device='cuda:1'), covar=tensor([0.0474, 0.2037, 0.0124, 0.0624, 0.0204, 0.0373, 0.0406, 0.2081], device='cuda:1'), in_proj_covar=tensor([0.0074, 0.0137, 0.0055, 0.0074, 0.0062, 0.0071, 0.0060, 0.0149], device='cuda:1'), out_proj_covar=tensor([4.9797e-05, 1.0278e-04, 3.5120e-05, 5.1023e-05, 4.0835e-05, 4.7463e-05, 4.3682e-05, 1.0921e-04], device='cuda:1') 2022-12-07 05:41:49,184 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:42:39,195 INFO [train.py:873] (1/4) Epoch 1, batch 4900, loss[loss=0.2617, simple_loss=0.255, pruned_loss=0.1342, over 14292.00 frames. ], tot_loss[loss=0.2818, simple_loss=0.2572, pruned_loss=0.1532, over 1972403.81 frames. ], batch size: 35, lr: 4.23e-02, grad_scale: 16.0 2022-12-07 05:42:44,141 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 3.107e+02 4.475e+02 6.071e+02 1.419e+03, threshold=8.951e+02, percent-clipped=8.0 2022-12-07 05:43:16,617 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:43:58,608 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:44:05,013 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 05:44:10,010 INFO [train.py:873] (1/4) Epoch 1, batch 5000, loss[loss=0.2858, simple_loss=0.2701, pruned_loss=0.1508, over 14317.00 frames. ], tot_loss[loss=0.2809, simple_loss=0.2571, pruned_loss=0.1524, over 2015470.10 frames. ], batch size: 28, lr: 4.20e-02, grad_scale: 16.0 2022-12-07 05:44:15,205 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 3.215e+02 4.595e+02 5.686e+02 1.097e+03, threshold=9.191e+02, percent-clipped=3.0 2022-12-07 05:44:50,583 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2808, 2.3943, 2.5650, 2.4700, 2.3001, 2.0838, 2.3122, 2.0000], device='cuda:1'), covar=tensor([0.0371, 0.0345, 0.0191, 0.0415, 0.0320, 0.1069, 0.0158, 0.1015], device='cuda:1'), in_proj_covar=tensor([0.0044, 0.0045, 0.0033, 0.0042, 0.0047, 0.0071, 0.0035, 0.0067], device='cuda:1'), out_proj_covar=tensor([3.4838e-05, 3.5573e-05, 2.5347e-05, 3.4391e-05, 3.6628e-05, 5.9755e-05, 2.4958e-05, 5.5050e-05], device='cuda:1') 2022-12-07 05:44:53,004 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5682, 2.3186, 2.4026, 2.4586, 2.3836, 2.5044, 2.4222, 2.3039], device='cuda:1'), covar=tensor([0.0341, 0.1691, 0.0172, 0.0599, 0.0221, 0.0394, 0.0423, 0.2022], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0147, 0.0060, 0.0080, 0.0068, 0.0076, 0.0067, 0.0164], device='cuda:1'), out_proj_covar=tensor([5.6449e-05, 1.1095e-04, 3.9833e-05, 5.7046e-05, 4.6662e-05, 5.2191e-05, 5.0463e-05, 1.2018e-04], device='cuda:1') 2022-12-07 05:45:22,386 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6936, 1.6768, 1.8480, 1.6996, 2.0641, 1.4706, 1.7659, 2.1462], device='cuda:1'), covar=tensor([0.2519, 0.1228, 0.1129, 0.0634, 0.0327, 0.0853, 0.0538, 0.0322], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0055, 0.0034, 0.0034, 0.0041, 0.0033, 0.0043, 0.0041], device='cuda:1'), out_proj_covar=tensor([1.1327e-04, 7.2270e-05, 5.0655e-05, 4.9982e-05, 5.1185e-05, 4.7862e-05, 5.8759e-05, 5.1141e-05], device='cuda:1') 2022-12-07 05:45:39,007 INFO [train.py:873] (1/4) Epoch 1, batch 5100, loss[loss=0.2732, simple_loss=0.2364, pruned_loss=0.155, over 4942.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.255, pruned_loss=0.1506, over 1989200.64 frames. ], batch size: 100, lr: 4.18e-02, grad_scale: 16.0 2022-12-07 05:45:43,999 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 3.085e+02 4.030e+02 5.035e+02 8.863e+02, threshold=8.060e+02, percent-clipped=0.0 2022-12-07 05:47:06,724 INFO [train.py:873] (1/4) Epoch 1, batch 5200, loss[loss=0.2564, simple_loss=0.2483, pruned_loss=0.1322, over 14457.00 frames. ], tot_loss[loss=0.2794, simple_loss=0.2558, pruned_loss=0.1515, over 1977695.39 frames. ], batch size: 51, lr: 4.16e-02, grad_scale: 16.0 2022-12-07 05:47:11,951 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 3.667e+02 4.771e+02 6.312e+02 1.162e+03, threshold=9.542e+02, percent-clipped=12.0 2022-12-07 05:47:30,547 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 05:48:18,903 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2058, 4.2107, 3.4800, 4.2529, 4.1952, 4.5648, 3.9547, 2.3839], device='cuda:1'), covar=tensor([0.0188, 0.0245, 0.1995, 0.0316, 0.0158, 0.0102, 0.0427, 0.2685], device='cuda:1'), in_proj_covar=tensor([0.0096, 0.0105, 0.0181, 0.0086, 0.0080, 0.0088, 0.0103, 0.0203], device='cuda:1'), out_proj_covar=tensor([5.4316e-05, 6.4403e-05, 1.1229e-04, 4.9499e-05, 4.5351e-05, 5.1348e-05, 6.2073e-05, 1.2523e-04], device='cuda:1') 2022-12-07 05:48:35,988 INFO [train.py:873] (1/4) Epoch 1, batch 5300, loss[loss=0.296, simple_loss=0.2751, pruned_loss=0.1584, over 13966.00 frames. ], tot_loss[loss=0.2767, simple_loss=0.2546, pruned_loss=0.1494, over 2007229.52 frames. ], batch size: 26, lr: 4.14e-02, grad_scale: 16.0 2022-12-07 05:48:40,962 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 3.215e+02 4.180e+02 5.101e+02 1.112e+03, threshold=8.360e+02, percent-clipped=0.0 2022-12-07 05:48:47,480 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3168, 1.8330, 2.0832, 1.5131, 1.9305, 1.7470, 1.3440, 2.2998], device='cuda:1'), covar=tensor([0.0213, 0.2320, 0.0400, 0.1131, 0.0674, 0.0528, 0.1516, 0.0442], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0092, 0.0043, 0.0066, 0.0042, 0.0047, 0.0041, 0.0047], device='cuda:1'), out_proj_covar=tensor([4.4733e-05, 1.0774e-04, 5.1890e-05, 7.9993e-05, 5.6564e-05, 6.0283e-05, 5.8274e-05, 5.5353e-05], device='cuda:1') 2022-12-07 05:50:04,682 INFO [train.py:873] (1/4) Epoch 1, batch 5400, loss[loss=0.2704, simple_loss=0.2364, pruned_loss=0.1522, over 5975.00 frames. ], tot_loss[loss=0.274, simple_loss=0.2534, pruned_loss=0.1473, over 2093494.97 frames. ], batch size: 100, lr: 4.12e-02, grad_scale: 16.0 2022-12-07 05:50:09,834 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 3.355e+02 4.207e+02 5.481e+02 1.330e+03, threshold=8.415e+02, percent-clipped=4.0 2022-12-07 05:50:39,706 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3798, 2.2409, 2.6486, 2.0744, 2.5758, 2.3925, 2.0573, 2.0398], device='cuda:1'), covar=tensor([0.0407, 0.1428, 0.0151, 0.0593, 0.0205, 0.0381, 0.0586, 0.1992], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0159, 0.0063, 0.0087, 0.0069, 0.0077, 0.0072, 0.0177], device='cuda:1'), out_proj_covar=tensor([5.9921e-05, 1.2015e-04, 4.2109e-05, 6.5671e-05, 4.8407e-05, 5.3928e-05, 5.6614e-05, 1.3065e-04], device='cuda:1') 2022-12-07 05:51:01,130 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9149, 4.4639, 4.3454, 4.8824, 4.7597, 4.8500, 4.5949, 4.3240], device='cuda:1'), covar=tensor([0.0224, 0.0420, 0.0332, 0.0374, 0.0374, 0.0298, 0.0442, 0.0349], device='cuda:1'), in_proj_covar=tensor([0.0045, 0.0070, 0.0056, 0.0052, 0.0054, 0.0061, 0.0073, 0.0068], device='cuda:1'), out_proj_covar=tensor([7.1378e-05, 9.9912e-05, 7.7523e-05, 8.0910e-05, 8.0994e-05, 9.3797e-05, 1.2053e-04, 1.0262e-04], device='cuda:1') 2022-12-07 05:51:01,197 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2326, 2.1554, 4.1442, 3.9019, 3.8927, 3.9136, 2.8990, 4.1463], device='cuda:1'), covar=tensor([0.1329, 0.1313, 0.0092, 0.0115, 0.0090, 0.0103, 0.0321, 0.0091], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0081, 0.0040, 0.0050, 0.0042, 0.0044, 0.0041, 0.0040], device='cuda:1'), out_proj_covar=tensor([1.1140e-04, 1.0978e-04, 5.5144e-05, 7.7237e-05, 5.5383e-05, 6.1735e-05, 6.2156e-05, 5.3502e-05], device='cuda:1') 2022-12-07 05:51:21,693 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0094, 3.7450, 3.8219, 4.1661, 4.0712, 4.1384, 4.1210, 3.7728], device='cuda:1'), covar=tensor([0.0420, 0.0813, 0.0380, 0.0488, 0.0433, 0.0415, 0.0448, 0.0477], device='cuda:1'), in_proj_covar=tensor([0.0045, 0.0073, 0.0056, 0.0053, 0.0055, 0.0061, 0.0074, 0.0070], device='cuda:1'), out_proj_covar=tensor([7.1502e-05, 1.0392e-04, 7.7834e-05, 8.2253e-05, 8.1992e-05, 9.3444e-05, 1.2125e-04, 1.0471e-04], device='cuda:1') 2022-12-07 05:51:33,375 INFO [train.py:873] (1/4) Epoch 1, batch 5500, loss[loss=0.2607, simple_loss=0.246, pruned_loss=0.1377, over 14249.00 frames. ], tot_loss[loss=0.2729, simple_loss=0.2526, pruned_loss=0.1466, over 2048881.29 frames. ], batch size: 69, lr: 4.10e-02, grad_scale: 16.0 2022-12-07 05:51:37,270 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2022-12-07 05:51:38,435 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 2.070e+02 3.601e+02 4.745e+02 6.042e+02 1.360e+03, threshold=9.490e+02, percent-clipped=8.0 2022-12-07 05:51:52,457 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2396, 2.2731, 4.1355, 3.0490, 3.8477, 2.3349, 4.2782, 4.0955], device='cuda:1'), covar=tensor([0.0283, 0.4691, 0.0342, 0.7867, 0.0138, 0.2923, 0.0332, 0.0155], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0239, 0.0137, 0.0330, 0.0103, 0.0263, 0.0156, 0.0111], device='cuda:1'), out_proj_covar=tensor([1.0938e-04, 1.9979e-04, 1.0447e-04, 2.7125e-04, 7.9157e-05, 2.1020e-04, 1.1995e-04, 8.2303e-05], device='cuda:1') 2022-12-07 05:52:39,243 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2022-12-07 05:53:01,582 INFO [train.py:873] (1/4) Epoch 1, batch 5600, loss[loss=0.2394, simple_loss=0.2137, pruned_loss=0.1326, over 3849.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.2519, pruned_loss=0.1467, over 2025498.97 frames. ], batch size: 100, lr: 4.08e-02, grad_scale: 16.0 2022-12-07 05:53:06,469 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 3.374e+02 4.886e+02 6.824e+02 1.449e+03, threshold=9.773e+02, percent-clipped=6.0 2022-12-07 05:53:06,699 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3330, 2.0626, 1.6114, 1.4912, 2.0620, 1.2806, 1.7119, 2.4824], device='cuda:1'), covar=tensor([0.0164, 0.1360, 0.0423, 0.0756, 0.0451, 0.0711, 0.0652, 0.0391], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0090, 0.0044, 0.0064, 0.0039, 0.0046, 0.0042, 0.0044], device='cuda:1'), out_proj_covar=tensor([4.4307e-05, 1.0977e-04, 5.5356e-05, 7.9904e-05, 5.3862e-05, 6.1893e-05, 6.2024e-05, 5.6067e-05], device='cuda:1') 2022-12-07 05:53:06,982 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 05:53:21,626 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 05:53:44,242 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:53:59,455 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0087, 3.8729, 3.7078, 4.0072, 4.0722, 4.0515, 3.9132, 3.6009], device='cuda:1'), covar=tensor([0.0319, 0.0388, 0.0333, 0.0397, 0.0252, 0.0321, 0.0506, 0.0456], device='cuda:1'), in_proj_covar=tensor([0.0047, 0.0072, 0.0058, 0.0054, 0.0056, 0.0064, 0.0078, 0.0072], device='cuda:1'), out_proj_covar=tensor([7.6965e-05, 1.0343e-04, 8.2302e-05, 8.4125e-05, 8.4176e-05, 1.0093e-04, 1.2859e-04, 1.1020e-04], device='cuda:1') 2022-12-07 05:54:29,451 INFO [train.py:873] (1/4) Epoch 1, batch 5700, loss[loss=0.2996, simple_loss=0.2712, pruned_loss=0.164, over 14114.00 frames. ], tot_loss[loss=0.2719, simple_loss=0.2511, pruned_loss=0.1463, over 2007060.19 frames. ], batch size: 29, lr: 4.06e-02, grad_scale: 16.0 2022-12-07 05:54:34,444 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 3.498e+02 4.626e+02 6.502e+02 1.133e+03, threshold=9.251e+02, percent-clipped=2.0 2022-12-07 05:54:38,246 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:54:52,454 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2022-12-07 05:55:09,046 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7818, 1.6943, 2.1987, 2.1484, 1.8930, 1.5904, 1.9399, 1.8451], device='cuda:1'), covar=tensor([0.0256, 0.0274, 0.0180, 0.0246, 0.0197, 0.0754, 0.0111, 0.0570], device='cuda:1'), in_proj_covar=tensor([0.0052, 0.0056, 0.0040, 0.0053, 0.0055, 0.0090, 0.0039, 0.0086], device='cuda:1'), out_proj_covar=tensor([4.3738e-05, 4.9358e-05, 3.4821e-05, 4.7484e-05, 4.6942e-05, 7.8682e-05, 3.0375e-05, 7.6938e-05], device='cuda:1') 2022-12-07 05:55:16,810 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:55:33,337 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 05:55:34,594 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8404, 1.3690, 3.7610, 3.5555, 3.3710, 3.1532, 2.5944, 3.9276], device='cuda:1'), covar=tensor([0.1704, 0.1825, 0.0149, 0.0128, 0.0117, 0.0218, 0.0342, 0.0089], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0089, 0.0044, 0.0054, 0.0046, 0.0047, 0.0045, 0.0042], device='cuda:1'), out_proj_covar=tensor([1.2601e-04, 1.2364e-04, 6.3552e-05, 8.5306e-05, 6.2605e-05, 6.8199e-05, 6.8910e-05, 5.7679e-05], device='cuda:1') 2022-12-07 05:55:34,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7625, 2.0136, 2.9217, 2.5741, 2.7033, 2.8462, 1.3414, 1.5102], device='cuda:1'), covar=tensor([0.0764, 0.0443, 0.0469, 0.0201, 0.0249, 0.0213, 0.1458, 0.0928], device='cuda:1'), in_proj_covar=tensor([0.0045, 0.0047, 0.0045, 0.0041, 0.0046, 0.0041, 0.0073, 0.0049], device='cuda:1'), out_proj_covar=tensor([5.0901e-05, 5.1182e-05, 5.0656e-05, 3.8458e-05, 4.3765e-05, 4.0435e-05, 8.3368e-05, 5.4360e-05], device='cuda:1') 2022-12-07 05:55:58,101 INFO [train.py:873] (1/4) Epoch 1, batch 5800, loss[loss=0.2712, simple_loss=0.2616, pruned_loss=0.1404, over 14285.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.2513, pruned_loss=0.1462, over 2055484.17 frames. ], batch size: 31, lr: 4.04e-02, grad_scale: 16.0 2022-12-07 05:56:03,252 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 3.658e+02 4.622e+02 6.453e+02 1.135e+03, threshold=9.244e+02, percent-clipped=6.0 2022-12-07 05:56:11,981 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:56:48,685 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:56:51,777 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 05:57:26,650 INFO [train.py:873] (1/4) Epoch 1, batch 5900, loss[loss=0.2598, simple_loss=0.2516, pruned_loss=0.134, over 14392.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.2511, pruned_loss=0.1462, over 2027378.37 frames. ], batch size: 41, lr: 4.02e-02, grad_scale: 16.0 2022-12-07 05:57:31,833 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.800e+02 3.450e+02 4.399e+02 5.936e+02 1.198e+03, threshold=8.798e+02, percent-clipped=6.0 2022-12-07 05:57:42,935 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:58:31,709 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9234, 1.2600, 1.6118, 1.2656, 1.5818, 0.8840, 1.6177, 2.0000], device='cuda:1'), covar=tensor([0.0320, 0.4135, 0.0610, 0.1277, 0.1140, 0.1131, 0.1199, 0.0438], device='cuda:1'), in_proj_covar=tensor([0.0037, 0.0102, 0.0047, 0.0069, 0.0045, 0.0049, 0.0045, 0.0045], device='cuda:1'), out_proj_covar=tensor([4.9502e-05, 1.2994e-04, 6.1213e-05, 9.0100e-05, 6.4552e-05, 6.8261e-05, 6.9848e-05, 6.0378e-05], device='cuda:1') 2022-12-07 05:58:44,825 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:58:52,950 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2022-12-07 05:58:54,699 INFO [train.py:873] (1/4) Epoch 1, batch 6000, loss[loss=0.2499, simple_loss=0.2247, pruned_loss=0.1375, over 4991.00 frames. ], tot_loss[loss=0.2708, simple_loss=0.2502, pruned_loss=0.1457, over 1955260.65 frames. ], batch size: 100, lr: 4.00e-02, grad_scale: 16.0 2022-12-07 05:58:54,699 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 05:59:02,829 INFO [train.py:905] (1/4) Epoch 1, validation: loss=0.159, simple_loss=0.1938, pruned_loss=0.06211, over 857387.00 frames. 2022-12-07 05:59:02,830 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17367MB 2022-12-07 05:59:07,193 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:59:07,937 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 3.896e+02 5.286e+02 6.584e+02 1.638e+03, threshold=1.057e+03, percent-clipped=9.0 2022-12-07 05:59:33,178 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7641, 3.4673, 3.5096, 3.3255, 3.7309, 3.7447, 3.8209, 3.6427], device='cuda:1'), covar=tensor([0.0401, 0.0555, 0.0517, 0.0592, 0.0334, 0.0312, 0.0451, 0.0703], device='cuda:1'), in_proj_covar=tensor([0.0094, 0.0087, 0.0113, 0.0079, 0.0096, 0.0092, 0.0099, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 05:59:39,680 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:59:46,826 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:29,015 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.17 vs. limit=5.0 2022-12-07 06:00:30,194 INFO [train.py:873] (1/4) Epoch 1, batch 6100, loss[loss=0.2967, simple_loss=0.2668, pruned_loss=0.1633, over 14509.00 frames. ], tot_loss[loss=0.2698, simple_loss=0.2498, pruned_loss=0.1449, over 1970705.57 frames. ], batch size: 34, lr: 3.98e-02, grad_scale: 16.0 2022-12-07 06:00:31,250 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 06:00:33,339 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:35,715 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 3.351e+02 5.105e+02 6.275e+02 1.538e+03, threshold=1.021e+03, percent-clipped=3.0 2022-12-07 06:00:39,119 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:39,441 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 06:01:00,468 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:31,128 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:53,414 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:58,128 INFO [train.py:873] (1/4) Epoch 1, batch 6200, loss[loss=0.303, simple_loss=0.2697, pruned_loss=0.1682, over 13537.00 frames. ], tot_loss[loss=0.2699, simple_loss=0.2498, pruned_loss=0.145, over 1934509.52 frames. ], batch size: 100, lr: 3.96e-02, grad_scale: 16.0 2022-12-07 06:02:00,064 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:03,437 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.672e+01 3.177e+02 4.421e+02 6.121e+02 1.475e+03, threshold=8.841e+02, percent-clipped=2.0 2022-12-07 06:02:07,201 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:09,644 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:12,717 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 06:02:20,163 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:02:23,966 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6320, 0.5775, 0.8911, 0.9861, 1.3624, 1.2145, 1.0501, 0.8641], device='cuda:1'), covar=tensor([0.0546, 0.0727, 0.0816, 0.0219, 0.0147, 0.0145, 0.0255, 0.0465], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0019, 0.0017, 0.0016, 0.0018, 0.0017, 0.0018], device='cuda:1'), out_proj_covar=tensor([2.0124e-05, 2.0591e-05, 2.5128e-05, 1.8590e-05, 1.9779e-05, 2.0463e-05, 2.3509e-05, 2.1511e-05], device='cuda:1') 2022-12-07 06:02:24,825 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:54,095 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:55,595 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:03:00,863 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:03:24,375 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.87 vs. limit=5.0 2022-12-07 06:03:26,922 INFO [train.py:873] (1/4) Epoch 1, batch 6300, loss[loss=0.2822, simple_loss=0.2465, pruned_loss=0.159, over 4929.00 frames. ], tot_loss[loss=0.2693, simple_loss=0.2495, pruned_loss=0.1446, over 1942472.39 frames. ], batch size: 100, lr: 3.94e-02, grad_scale: 16.0 2022-12-07 06:03:31,268 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:03:31,914 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 3.956e+02 5.182e+02 6.845e+02 1.642e+03, threshold=1.036e+03, percent-clipped=11.0 2022-12-07 06:03:49,314 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:04:04,876 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2879, 3.1196, 3.1525, 2.9511, 3.2516, 3.1325, 3.2586, 3.2179], device='cuda:1'), covar=tensor([0.0314, 0.0417, 0.0385, 0.0543, 0.0258, 0.0310, 0.0448, 0.0512], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0088, 0.0106, 0.0071, 0.0092, 0.0090, 0.0096, 0.0081], device='cuda:1'), out_proj_covar=tensor([1.0955e-04, 1.1412e-04, 1.3326e-04, 9.1150e-05, 1.1321e-04, 1.1043e-04, 1.4007e-04, 1.0752e-04], device='cuda:1') 2022-12-07 06:04:05,599 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:04:12,438 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:04:33,263 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0458, 2.1368, 1.7519, 2.3657, 2.1960, 2.1910, 1.8449, 1.9520], device='cuda:1'), covar=tensor([0.0194, 0.0450, 0.1135, 0.0131, 0.0165, 0.0194, 0.0326, 0.0900], device='cuda:1'), in_proj_covar=tensor([0.0110, 0.0126, 0.0204, 0.0100, 0.0093, 0.0103, 0.0118, 0.0238], device='cuda:1'), out_proj_covar=tensor([6.5215e-05, 8.0531e-05, 1.2743e-04, 5.9534e-05, 5.8811e-05, 6.5093e-05, 7.4996e-05, 1.4687e-04], device='cuda:1') 2022-12-07 06:04:41,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4970, 1.3603, 1.9542, 1.7634, 1.7365, 1.6393, 0.9923, 1.3550], device='cuda:1'), covar=tensor([0.0312, 0.0453, 0.0153, 0.0198, 0.0222, 0.0198, 0.0762, 0.0515], device='cuda:1'), in_proj_covar=tensor([0.0044, 0.0046, 0.0042, 0.0042, 0.0046, 0.0042, 0.0078, 0.0050], device='cuda:1'), out_proj_covar=tensor([5.3130e-05, 5.2547e-05, 4.8865e-05, 4.0665e-05, 4.7506e-05, 4.3501e-05, 9.2001e-05, 5.9112e-05], device='cuda:1') 2022-12-07 06:04:47,092 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0564, 2.0454, 1.8635, 1.9753, 1.8617, 1.8800, 2.1239, 2.0830], device='cuda:1'), covar=tensor([0.1059, 0.0854, 0.1271, 0.1051, 0.1071, 0.0667, 0.1000, 0.0899], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0062, 0.0085, 0.0081, 0.0082, 0.0055, 0.0080, 0.0079], device='cuda:1'), out_proj_covar=tensor([1.1284e-04, 9.6220e-05, 1.2350e-04, 1.1818e-04, 1.1799e-04, 7.9425e-05, 1.2183e-04, 1.1697e-04], device='cuda:1') 2022-12-07 06:04:51,691 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:04:53,610 INFO [train.py:873] (1/4) Epoch 1, batch 6400, loss[loss=0.2858, simple_loss=0.2219, pruned_loss=0.1749, over 1196.00 frames. ], tot_loss[loss=0.2681, simple_loss=0.2489, pruned_loss=0.1436, over 1946251.09 frames. ], batch size: 100, lr: 3.92e-02, grad_scale: 8.0 2022-12-07 06:05:00,167 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 3.375e+02 4.095e+02 5.340e+02 1.039e+03, threshold=8.189e+02, percent-clipped=1.0 2022-12-07 06:05:03,086 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:44,993 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:47,717 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:49,320 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 06:05:52,063 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:12,998 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:21,241 INFO [train.py:873] (1/4) Epoch 1, batch 6500, loss[loss=0.2637, simple_loss=0.2502, pruned_loss=0.1386, over 14576.00 frames. ], tot_loss[loss=0.268, simple_loss=0.2488, pruned_loss=0.1436, over 1928819.24 frames. ], batch size: 49, lr: 3.90e-02, grad_scale: 8.0 2022-12-07 06:06:27,500 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.203e+01 3.294e+02 4.343e+02 5.711e+02 1.098e+03, threshold=8.686e+02, percent-clipped=4.0 2022-12-07 06:06:32,759 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:40,820 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:06:43,794 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:45,586 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:48,920 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8585, 1.3454, 1.3977, 0.9254, 1.4869, 1.1985, 1.3166, 1.5771], device='cuda:1'), covar=tensor([0.0193, 0.1859, 0.0394, 0.1009, 0.0448, 0.0446, 0.0790, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0041, 0.0107, 0.0051, 0.0074, 0.0045, 0.0049, 0.0047, 0.0047], device='cuda:1'), out_proj_covar=tensor([5.7369e-05, 1.4248e-04, 6.8619e-05, 1.0190e-04, 6.8281e-05, 7.2470e-05, 7.6336e-05, 6.5939e-05], device='cuda:1') 2022-12-07 06:07:07,058 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 06:07:11,167 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1582, 3.4184, 4.0710, 3.2066, 3.9450, 4.0103, 1.8525, 4.0440], device='cuda:1'), covar=tensor([0.0161, 0.0439, 0.0298, 0.0501, 0.0227, 0.0180, 0.2461, 0.0166], device='cuda:1'), in_proj_covar=tensor([0.0069, 0.0066, 0.0071, 0.0057, 0.0088, 0.0060, 0.0107, 0.0081], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 06:07:12,793 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:15,385 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:20,395 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:40,899 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.61 vs. limit=5.0 2022-12-07 06:07:50,329 INFO [train.py:873] (1/4) Epoch 1, batch 6600, loss[loss=0.2695, simple_loss=0.2523, pruned_loss=0.1434, over 6932.00 frames. ], tot_loss[loss=0.2669, simple_loss=0.2481, pruned_loss=0.1429, over 1955145.00 frames. ], batch size: 100, lr: 3.89e-02, grad_scale: 8.0 2022-12-07 06:07:54,339 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2087, 3.7335, 3.9892, 4.1521, 4.0289, 4.0760, 3.9637, 3.5895], device='cuda:1'), covar=tensor([0.0565, 0.1111, 0.0560, 0.0808, 0.0862, 0.0715, 0.1008, 0.0990], device='cuda:1'), in_proj_covar=tensor([0.0050, 0.0081, 0.0063, 0.0062, 0.0061, 0.0070, 0.0086, 0.0082], device='cuda:1'), out_proj_covar=tensor([8.1741e-05, 1.2293e-04, 9.0206e-05, 9.8182e-05, 9.8308e-05, 1.1327e-04, 1.4730e-04, 1.3002e-04], device='cuda:1') 2022-12-07 06:07:56,667 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 3.480e+02 4.603e+02 5.913e+02 1.121e+03, threshold=9.206e+02, percent-clipped=7.0 2022-12-07 06:08:04,775 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=4.57 vs. limit=2.0 2022-12-07 06:08:08,521 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:08:14,775 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:08:23,013 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0919, 1.8102, 3.8074, 2.4581, 3.8318, 3.7401, 2.9365, 4.2573], device='cuda:1'), covar=tensor([0.0207, 0.2391, 0.0303, 0.1654, 0.0248, 0.0254, 0.0573, 0.0167], device='cuda:1'), in_proj_covar=tensor([0.0059, 0.0099, 0.0054, 0.0100, 0.0073, 0.0057, 0.0056, 0.0056], device='cuda:1'), out_proj_covar=tensor([9.7481e-05, 1.5396e-04, 1.0010e-04, 1.5558e-04, 1.2627e-04, 9.9644e-05, 9.8165e-05, 9.6916e-05], device='cuda:1') 2022-12-07 06:08:30,324 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:08:39,302 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 06:08:52,358 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:09:09,278 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:12,787 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:17,189 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:18,766 INFO [train.py:873] (1/4) Epoch 1, batch 6700, loss[loss=0.3015, simple_loss=0.2716, pruned_loss=0.1657, over 14261.00 frames. ], tot_loss[loss=0.2672, simple_loss=0.2479, pruned_loss=0.1433, over 1929940.26 frames. ], batch size: 80, lr: 3.87e-02, grad_scale: 8.0 2022-12-07 06:09:22,293 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8347, 4.4593, 4.2822, 4.8016, 4.6355, 4.5894, 4.7915, 4.3444], device='cuda:1'), covar=tensor([0.0224, 0.0627, 0.0309, 0.0377, 0.0337, 0.0328, 0.0439, 0.0376], device='cuda:1'), in_proj_covar=tensor([0.0051, 0.0082, 0.0063, 0.0062, 0.0063, 0.0070, 0.0085, 0.0080], device='cuda:1'), out_proj_covar=tensor([8.3977e-05, 1.2416e-04, 9.1667e-05, 9.7655e-05, 1.0189e-04, 1.1502e-04, 1.4561e-04, 1.2737e-04], device='cuda:1') 2022-12-07 06:09:24,613 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 3.291e+02 4.382e+02 6.181e+02 1.245e+03, threshold=8.764e+02, percent-clipped=7.0 2022-12-07 06:09:58,744 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:10:12,159 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.07 vs. limit=2.0 2022-12-07 06:10:37,323 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:10:45,586 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1999, 2.9891, 2.9727, 3.3281, 2.9942, 2.5614, 3.3363, 3.2787], device='cuda:1'), covar=tensor([0.0863, 0.0848, 0.0923, 0.0742, 0.0849, 0.0981, 0.0876, 0.0996], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0062, 0.0087, 0.0078, 0.0083, 0.0057, 0.0077, 0.0081], device='cuda:1'), out_proj_covar=tensor([1.1511e-04, 9.7496e-05, 1.2730e-04, 1.1610e-04, 1.2097e-04, 8.2804e-05, 1.1778e-04, 1.2230e-04], device='cuda:1') 2022-12-07 06:10:46,418 INFO [train.py:873] (1/4) Epoch 1, batch 6800, loss[loss=0.27, simple_loss=0.2381, pruned_loss=0.1509, over 4942.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.2476, pruned_loss=0.1429, over 1925712.79 frames. ], batch size: 100, lr: 3.85e-02, grad_scale: 8.0 2022-12-07 06:10:52,797 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.483e+01 3.173e+02 4.494e+02 5.972e+02 9.591e+02, threshold=8.989e+02, percent-clipped=5.0 2022-12-07 06:11:01,540 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:11:05,695 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:08,562 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:19,529 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:37,234 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:37,244 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8444, 1.5739, 2.6996, 2.2550, 2.3586, 1.9280, 1.0807, 1.8993], device='cuda:1'), covar=tensor([0.0639, 0.0753, 0.0229, 0.0240, 0.0280, 0.0449, 0.1428, 0.0728], device='cuda:1'), in_proj_covar=tensor([0.0049, 0.0051, 0.0049, 0.0050, 0.0051, 0.0050, 0.0088, 0.0054], device='cuda:1'), out_proj_covar=tensor([6.1126e-05, 6.0480e-05, 5.9341e-05, 5.1086e-05, 5.3613e-05, 5.5624e-05, 1.0518e-04, 6.5817e-05], device='cuda:1') 2022-12-07 06:11:39,241 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.11 vs. limit=5.0 2022-12-07 06:11:44,253 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:49,844 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:53,686 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:12:13,453 INFO [train.py:873] (1/4) Epoch 1, batch 6900, loss[loss=0.2889, simple_loss=0.2613, pruned_loss=0.1582, over 14281.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.2475, pruned_loss=0.1435, over 1923017.49 frames. ], batch size: 60, lr: 3.83e-02, grad_scale: 8.0 2022-12-07 06:12:18,835 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:19,586 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 3.106e+02 4.451e+02 6.203e+02 1.044e+03, threshold=8.902e+02, percent-clipped=8.0 2022-12-07 06:12:20,668 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:23,122 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5187, 3.2208, 3.1121, 3.6135, 3.3213, 2.7839, 3.6562, 3.6615], device='cuda:1'), covar=tensor([0.0911, 0.0840, 0.0942, 0.0966, 0.0676, 0.0749, 0.0815, 0.0921], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0068, 0.0088, 0.0083, 0.0084, 0.0058, 0.0079, 0.0084], device='cuda:1'), out_proj_covar=tensor([1.2012e-04, 1.0560e-04, 1.2857e-04, 1.2362e-04, 1.2331e-04, 8.3650e-05, 1.2261e-04, 1.2764e-04], device='cuda:1') 2022-12-07 06:12:25,623 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:31,976 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:46,783 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:13:04,896 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 06:13:13,996 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:14,083 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:27,125 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:41,156 INFO [train.py:873] (1/4) Epoch 1, batch 7000, loss[loss=0.3052, simple_loss=0.2665, pruned_loss=0.172, over 8569.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.2467, pruned_loss=0.1433, over 1840812.11 frames. ], batch size: 100, lr: 3.81e-02, grad_scale: 8.0 2022-12-07 06:13:47,947 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 2.871e+02 4.067e+02 4.755e+02 1.195e+03, threshold=8.134e+02, percent-clipped=1.0 2022-12-07 06:14:21,571 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4450, 1.5746, 2.4055, 1.6265, 2.4277, 2.4835, 1.7391, 2.5434], device='cuda:1'), covar=tensor([0.0182, 0.1751, 0.0366, 0.1430, 0.0244, 0.0229, 0.0704, 0.0232], device='cuda:1'), in_proj_covar=tensor([0.0058, 0.0101, 0.0056, 0.0102, 0.0074, 0.0057, 0.0059, 0.0058], device='cuda:1'), out_proj_covar=tensor([9.9872e-05, 1.6288e-04, 1.0851e-04, 1.6299e-04, 1.3289e-04, 1.0256e-04, 1.0755e-04, 1.0482e-04], device='cuda:1') 2022-12-07 06:14:23,444 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:15:09,089 INFO [train.py:873] (1/4) Epoch 1, batch 7100, loss[loss=0.2503, simple_loss=0.2098, pruned_loss=0.1454, over 2671.00 frames. ], tot_loss[loss=0.2656, simple_loss=0.2466, pruned_loss=0.1423, over 1855992.15 frames. ], batch size: 100, lr: 3.79e-02, grad_scale: 8.0 2022-12-07 06:15:15,338 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 3.113e+02 4.115e+02 5.668e+02 1.201e+03, threshold=8.229e+02, percent-clipped=4.0 2022-12-07 06:15:16,475 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:15:16,730 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-07 06:15:18,427 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 06:15:23,777 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:15:26,979 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.41 vs. limit=2.0 2022-12-07 06:15:28,184 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:03,204 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:05,593 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:09,758 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:23,928 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8451, 3.5147, 3.4384, 3.9173, 3.5799, 2.8498, 3.9546, 3.9264], device='cuda:1'), covar=tensor([0.0538, 0.0699, 0.0749, 0.0548, 0.0738, 0.0740, 0.0555, 0.0683], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0066, 0.0087, 0.0079, 0.0085, 0.0056, 0.0078, 0.0083], device='cuda:1'), out_proj_covar=tensor([1.1825e-04, 1.0563e-04, 1.2969e-04, 1.1991e-04, 1.2651e-04, 8.2713e-05, 1.2194e-04, 1.2587e-04], device='cuda:1') 2022-12-07 06:16:35,874 INFO [train.py:873] (1/4) Epoch 1, batch 7200, loss[loss=0.2998, simple_loss=0.2631, pruned_loss=0.1682, over 10352.00 frames. ], tot_loss[loss=0.2655, simple_loss=0.2465, pruned_loss=0.1422, over 1896205.44 frames. ], batch size: 100, lr: 3.78e-02, grad_scale: 8.0 2022-12-07 06:16:42,299 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.199e+02 3.143e+02 4.183e+02 5.672e+02 1.523e+03, threshold=8.365e+02, percent-clipped=9.0 2022-12-07 06:16:56,276 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 06:17:04,762 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:17:18,800 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.79 vs. limit=2.0 2022-12-07 06:17:32,304 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:17:33,607 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.05 vs. limit=2.0 2022-12-07 06:17:41,197 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:17:49,222 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8510, 2.4962, 2.9850, 2.4635, 2.8146, 3.1897, 3.0628, 2.6645], device='cuda:1'), covar=tensor([0.0342, 0.1675, 0.0143, 0.0905, 0.0173, 0.0272, 0.0352, 0.1481], device='cuda:1'), in_proj_covar=tensor([0.0110, 0.0208, 0.0080, 0.0128, 0.0083, 0.0105, 0.0085, 0.0232], device='cuda:1'), out_proj_covar=tensor([8.6737e-05, 1.6423e-04, 5.9792e-05, 1.0271e-04, 6.3408e-05, 8.1687e-05, 7.9172e-05, 1.7928e-04], device='cuda:1') 2022-12-07 06:17:50,856 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:18:04,545 INFO [train.py:873] (1/4) Epoch 1, batch 7300, loss[loss=0.2143, simple_loss=0.1848, pruned_loss=0.1219, over 1246.00 frames. ], tot_loss[loss=0.263, simple_loss=0.2448, pruned_loss=0.1406, over 1863958.24 frames. ], batch size: 100, lr: 3.76e-02, grad_scale: 8.0 2022-12-07 06:18:10,449 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 3.489e+02 4.612e+02 5.728e+02 1.039e+03, threshold=9.225e+02, percent-clipped=2.0 2022-12-07 06:18:24,769 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2184, 2.9240, 3.4038, 3.0680, 3.2142, 2.2695, 2.6770, 2.0538], device='cuda:1'), covar=tensor([0.0184, 0.0280, 0.0120, 0.0279, 0.0180, 0.0888, 0.0079, 0.0987], device='cuda:1'), in_proj_covar=tensor([0.0066, 0.0076, 0.0054, 0.0072, 0.0076, 0.0116, 0.0048, 0.0119], device='cuda:1'), out_proj_covar=tensor([6.4100e-05, 7.7530e-05, 5.5834e-05, 7.6812e-05, 7.9930e-05, 1.1690e-04, 4.4410e-05, 1.1877e-04], device='cuda:1') 2022-12-07 06:18:32,904 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:18:35,678 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:18:50,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7046, 0.7005, 0.7938, 0.5422, 0.5492, 0.7759, 0.7066, 0.5870], device='cuda:1'), covar=tensor([0.0036, 0.0024, 0.0047, 0.0039, 0.0090, 0.0040, 0.0022, 0.0105], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0016, 0.0019, 0.0018, 0.0018, 0.0018, 0.0014, 0.0017], device='cuda:1'), out_proj_covar=tensor([2.3518e-05, 2.1057e-05, 2.7737e-05, 2.2416e-05, 2.4700e-05, 2.1961e-05, 2.1219e-05, 2.3274e-05], device='cuda:1') 2022-12-07 06:19:33,641 INFO [train.py:873] (1/4) Epoch 1, batch 7400, loss[loss=0.2247, simple_loss=0.209, pruned_loss=0.1202, over 4986.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.2444, pruned_loss=0.1401, over 1866665.70 frames. ], batch size: 100, lr: 3.74e-02, grad_scale: 8.0 2022-12-07 06:19:36,755 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:19:39,336 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:19:39,942 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 3.232e+02 4.380e+02 5.803e+02 1.577e+03, threshold=8.760e+02, percent-clipped=3.0 2022-12-07 06:19:51,399 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8526, 0.7408, 0.6644, 0.9777, 0.9753, 0.5826, 0.7797, 0.7292], device='cuda:1'), covar=tensor([0.0458, 0.0580, 0.0288, 0.0326, 0.0309, 0.0341, 0.0334, 0.0248], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0024, 0.0026, 0.0021, 0.0022, 0.0027, 0.0019, 0.0022], device='cuda:1'), out_proj_covar=tensor([3.2590e-05, 3.9377e-05, 3.8046e-05, 3.5449e-05, 3.1641e-05, 4.2144e-05, 3.0463e-05, 3.1648e-05], device='cuda:1') 2022-12-07 06:20:12,614 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8560, 1.4431, 2.0409, 1.5592, 1.8808, 1.9232, 1.4099, 2.0042], device='cuda:1'), covar=tensor([0.0167, 0.0771, 0.0193, 0.0947, 0.0242, 0.0251, 0.0521, 0.0191], device='cuda:1'), in_proj_covar=tensor([0.0061, 0.0106, 0.0059, 0.0110, 0.0078, 0.0064, 0.0059, 0.0062], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 06:20:32,081 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:20:34,549 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2022-12-07 06:21:00,784 INFO [train.py:873] (1/4) Epoch 1, batch 7500, loss[loss=0.2669, simple_loss=0.2454, pruned_loss=0.1442, over 14082.00 frames. ], tot_loss[loss=0.2617, simple_loss=0.2442, pruned_loss=0.1396, over 1884481.90 frames. ], batch size: 29, lr: 3.72e-02, grad_scale: 8.0 2022-12-07 06:21:06,494 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 3.575e+02 4.286e+02 5.078e+02 8.997e+02, threshold=8.573e+02, percent-clipped=1.0 2022-12-07 06:21:16,624 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:21:29,238 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:22:29,584 INFO [train.py:873] (1/4) Epoch 2, batch 0, loss[loss=0.3618, simple_loss=0.3231, pruned_loss=0.2002, over 14209.00 frames. ], tot_loss[loss=0.3618, simple_loss=0.3231, pruned_loss=0.2002, over 14209.00 frames. ], batch size: 35, lr: 3.64e-02, grad_scale: 8.0 2022-12-07 06:22:29,584 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 06:22:36,878 INFO [train.py:905] (1/4) Epoch 2, validation: loss=0.201, simple_loss=0.225, pruned_loss=0.08852, over 857387.00 frames. 2022-12-07 06:22:36,878 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 06:22:38,750 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:22:54,112 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:23:16,974 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 3.396e+01 2.991e+02 4.594e+02 6.425e+02 1.765e+03, threshold=9.187e+02, percent-clipped=13.0 2022-12-07 06:23:19,561 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3626, 3.9404, 4.1965, 4.5090, 4.2120, 3.4094, 4.5935, 4.5888], device='cuda:1'), covar=tensor([0.0840, 0.0603, 0.0631, 0.0647, 0.0773, 0.0616, 0.0686, 0.0705], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0067, 0.0092, 0.0083, 0.0090, 0.0060, 0.0081, 0.0088], device='cuda:1'), out_proj_covar=tensor([1.3289e-04, 1.0612e-04, 1.3858e-04, 1.2877e-04, 1.3519e-04, 9.1601e-05, 1.2906e-04, 1.3558e-04], device='cuda:1') 2022-12-07 06:23:21,541 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:23:31,664 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:23:37,426 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:23:41,718 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5093, 1.9813, 1.2888, 0.9028, 1.8256, 1.8547, 1.2044, 1.8684], device='cuda:1'), covar=tensor([0.0223, 0.2270, 0.0962, 0.1541, 0.0427, 0.0495, 0.1796, 0.0502], device='cuda:1'), in_proj_covar=tensor([0.0042, 0.0110, 0.0055, 0.0079, 0.0047, 0.0051, 0.0050, 0.0048], device='cuda:1'), out_proj_covar=tensor([6.5996e-05, 1.6393e-04, 8.6625e-05, 1.1762e-04, 8.0963e-05, 8.5043e-05, 8.8639e-05, 7.6100e-05], device='cuda:1') 2022-12-07 06:23:47,078 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 06:24:05,652 INFO [train.py:873] (1/4) Epoch 2, batch 100, loss[loss=0.2348, simple_loss=0.2356, pruned_loss=0.117, over 14314.00 frames. ], tot_loss[loss=0.2637, simple_loss=0.247, pruned_loss=0.1402, over 845063.91 frames. ], batch size: 28, lr: 3.62e-02, grad_scale: 8.0 2022-12-07 06:24:19,208 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9811, 1.7415, 1.4240, 1.0462, 1.2643, 1.2808, 1.8896, 0.9137], device='cuda:1'), covar=tensor([0.0925, 0.0595, 0.0549, 0.0347, 0.0507, 0.0356, 0.0238, 0.0349], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0017, 0.0017, 0.0019, 0.0019, 0.0019, 0.0017, 0.0018], device='cuda:1'), out_proj_covar=tensor([2.2461e-05, 2.2757e-05, 2.6805e-05, 2.4244e-05, 2.5231e-05, 2.2895e-05, 2.5867e-05, 2.4192e-05], device='cuda:1') 2022-12-07 06:24:20,833 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1796, 2.1292, 3.9031, 2.5079, 4.0677, 3.9171, 3.8017, 4.5594], device='cuda:1'), covar=tensor([0.0289, 0.2345, 0.0373, 0.1990, 0.0253, 0.0310, 0.0316, 0.0233], device='cuda:1'), in_proj_covar=tensor([0.0066, 0.0108, 0.0062, 0.0112, 0.0081, 0.0066, 0.0064, 0.0062], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 06:24:24,538 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:24:25,486 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2627, 2.2266, 2.6328, 2.5370, 2.3184, 2.1050, 2.6337, 2.0263], device='cuda:1'), covar=tensor([0.0190, 0.0258, 0.0136, 0.0260, 0.0202, 0.0665, 0.0083, 0.0698], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0083, 0.0056, 0.0083, 0.0084, 0.0129, 0.0053, 0.0130], device='cuda:1'), out_proj_covar=tensor([7.2298e-05, 8.6781e-05, 6.1636e-05, 9.0545e-05, 9.1609e-05, 1.3581e-04, 5.1765e-05, 1.3315e-04], device='cuda:1') 2022-12-07 06:24:27,656 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 06:24:31,235 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.46 vs. limit=2.0 2022-12-07 06:24:41,615 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:24:44,979 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 3.684e+02 4.699e+02 5.885e+02 1.470e+03, threshold=9.397e+02, percent-clipped=3.0 2022-12-07 06:25:16,169 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:25:23,874 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:25:32,821 INFO [train.py:873] (1/4) Epoch 2, batch 200, loss[loss=0.2572, simple_loss=0.2499, pruned_loss=0.1322, over 14138.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.2443, pruned_loss=0.1377, over 1289337.50 frames. ], batch size: 84, lr: 3.61e-02, grad_scale: 8.0 2022-12-07 06:25:33,258 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:06,975 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:08,681 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:12,300 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 3.123e+02 4.049e+02 5.757e+02 1.056e+03, threshold=8.099e+02, percent-clipped=3.0 2022-12-07 06:26:22,312 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:26:43,482 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 06:26:59,714 INFO [train.py:873] (1/4) Epoch 2, batch 300, loss[loss=0.229, simple_loss=0.2273, pruned_loss=0.1153, over 13919.00 frames. ], tot_loss[loss=0.2581, simple_loss=0.2426, pruned_loss=0.1368, over 1505248.81 frames. ], batch size: 19, lr: 3.59e-02, grad_scale: 8.0 2022-12-07 06:26:59,908 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:27:03,174 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:27:38,763 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 3.418e+02 4.416e+02 5.390e+02 1.169e+03, threshold=8.833e+02, percent-clipped=6.0 2022-12-07 06:27:53,814 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-07 06:27:58,882 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:28:22,370 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 06:28:26,417 INFO [train.py:873] (1/4) Epoch 2, batch 400, loss[loss=0.2976, simple_loss=0.265, pruned_loss=0.1651, over 11207.00 frames. ], tot_loss[loss=0.2576, simple_loss=0.2425, pruned_loss=0.1363, over 1685334.30 frames. ], batch size: 100, lr: 3.58e-02, grad_scale: 8.0 2022-12-07 06:28:40,961 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:28:41,837 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:29:05,269 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.294e+01 3.140e+02 4.082e+02 5.332e+02 1.723e+03, threshold=8.164e+02, percent-clipped=4.0 2022-12-07 06:29:20,517 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0165, 2.0015, 1.8406, 1.7786, 2.0054, 1.9165, 1.9549, 1.9770], device='cuda:1'), covar=tensor([0.0562, 0.0556, 0.0948, 0.0833, 0.0432, 0.0447, 0.0933, 0.0636], device='cuda:1'), in_proj_covar=tensor([0.0103, 0.0098, 0.0122, 0.0088, 0.0103, 0.0106, 0.0121, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 06:29:24,724 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2022-12-07 06:29:53,231 INFO [train.py:873] (1/4) Epoch 2, batch 500, loss[loss=0.262, simple_loss=0.2453, pruned_loss=0.1394, over 14381.00 frames. ], tot_loss[loss=0.2593, simple_loss=0.2437, pruned_loss=0.1375, over 1807889.85 frames. ], batch size: 55, lr: 3.56e-02, grad_scale: 8.0 2022-12-07 06:29:53,380 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:25,042 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:27,245 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 06:30:32,479 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.590e+01 3.381e+02 4.581e+02 6.129e+02 1.327e+03, threshold=9.162e+02, percent-clipped=13.0 2022-12-07 06:30:35,065 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:40,317 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4750, 3.1414, 3.1605, 3.5071, 3.2253, 2.8292, 3.5208, 3.4762], device='cuda:1'), covar=tensor([0.0722, 0.0658, 0.0798, 0.0703, 0.0637, 0.0747, 0.0729, 0.0899], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0067, 0.0093, 0.0088, 0.0088, 0.0060, 0.0082, 0.0087], device='cuda:1'), out_proj_covar=tensor([1.3200e-04, 1.0683e-04, 1.4427e-04, 1.3699e-04, 1.3531e-04, 9.2429e-05, 1.3097e-04, 1.3529e-04], device='cuda:1') 2022-12-07 06:30:57,791 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3384, 1.9675, 3.4811, 2.7056, 3.3818, 2.1443, 3.3227, 3.2896], device='cuda:1'), covar=tensor([0.0259, 0.4423, 0.0197, 0.6497, 0.0145, 0.2828, 0.0432, 0.0206], device='cuda:1'), in_proj_covar=tensor([0.0159, 0.0265, 0.0136, 0.0366, 0.0112, 0.0284, 0.0180, 0.0127], device='cuda:1'), out_proj_covar=tensor([1.3884e-04, 2.4302e-04, 1.2081e-04, 3.1669e-04, 9.9834e-05, 2.4489e-04, 1.5380e-04, 1.1312e-04], device='cuda:1') 2022-12-07 06:31:15,354 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:31:19,467 INFO [train.py:873] (1/4) Epoch 2, batch 600, loss[loss=0.1683, simple_loss=0.1481, pruned_loss=0.09419, over 1240.00 frames. ], tot_loss[loss=0.259, simple_loss=0.2431, pruned_loss=0.1374, over 1891402.84 frames. ], batch size: 100, lr: 3.54e-02, grad_scale: 8.0 2022-12-07 06:31:47,804 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7234, 4.4303, 4.9418, 3.8657, 4.7747, 4.7549, 2.5477, 4.5956], device='cuda:1'), covar=tensor([0.0186, 0.0271, 0.0326, 0.0400, 0.0191, 0.0137, 0.2634, 0.0193], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0075, 0.0081, 0.0065, 0.0103, 0.0069, 0.0122, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 06:31:56,024 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=29.71 vs. limit=5.0 2022-12-07 06:31:59,046 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 3.066e+02 3.782e+02 5.192e+02 1.492e+03, threshold=7.564e+02, percent-clipped=6.0 2022-12-07 06:32:21,131 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3035, 4.7715, 4.9177, 5.4137, 5.2670, 4.7816, 5.5799, 5.4981], device='cuda:1'), covar=tensor([0.0504, 0.0466, 0.0570, 0.0494, 0.0380, 0.0271, 0.0390, 0.0497], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0065, 0.0092, 0.0086, 0.0088, 0.0059, 0.0080, 0.0087], device='cuda:1'), out_proj_covar=tensor([1.2804e-04, 1.0374e-04, 1.4273e-04, 1.3342e-04, 1.3595e-04, 9.1159e-05, 1.2907e-04, 1.3594e-04], device='cuda:1') 2022-12-07 06:32:47,259 INFO [train.py:873] (1/4) Epoch 2, batch 700, loss[loss=0.2255, simple_loss=0.1929, pruned_loss=0.1291, over 2565.00 frames. ], tot_loss[loss=0.2563, simple_loss=0.241, pruned_loss=0.1358, over 1890624.35 frames. ], batch size: 100, lr: 3.53e-02, grad_scale: 8.0 2022-12-07 06:32:48,774 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-07 06:32:49,141 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:32:52,667 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 06:33:01,837 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:33:21,931 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5418, 1.3105, 1.2267, 0.7874, 1.4010, 1.0944, 1.3257, 1.4407], device='cuda:1'), covar=tensor([0.0199, 0.1677, 0.0409, 0.0826, 0.0424, 0.0486, 0.0380, 0.0337], device='cuda:1'), in_proj_covar=tensor([0.0042, 0.0112, 0.0054, 0.0076, 0.0051, 0.0050, 0.0048, 0.0047], device='cuda:1'), out_proj_covar=tensor([6.8160e-05, 1.7457e-04, 8.9652e-05, 1.2264e-04, 9.1895e-05, 8.8645e-05, 8.8110e-05, 7.9178e-05], device='cuda:1') 2022-12-07 06:33:26,032 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 3.101e+02 3.944e+02 5.331e+02 1.080e+03, threshold=7.888e+02, percent-clipped=5.0 2022-12-07 06:33:27,064 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8169, 1.5167, 3.5518, 3.2839, 3.2180, 3.4042, 2.6262, 3.6073], device='cuda:1'), covar=tensor([0.1611, 0.1651, 0.0100, 0.0164, 0.0137, 0.0132, 0.0348, 0.0105], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0109, 0.0053, 0.0071, 0.0060, 0.0060, 0.0053, 0.0052], device='cuda:1'), out_proj_covar=tensor([1.8113e-04, 1.7736e-04, 9.1420e-05, 1.3660e-04, 1.0159e-04, 1.0738e-04, 1.0103e-04, 8.9453e-05], device='cuda:1') 2022-12-07 06:33:35,563 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1104, 0.9802, 1.2022, 1.0164, 0.9414, 1.0108, 0.8851, 0.9284], device='cuda:1'), covar=tensor([0.1133, 0.0764, 0.0293, 0.0251, 0.0469, 0.0248, 0.0723, 0.0341], device='cuda:1'), in_proj_covar=tensor([0.0112, 0.0052, 0.0038, 0.0041, 0.0049, 0.0039, 0.0048, 0.0049], device='cuda:1'), out_proj_covar=tensor([1.9965e-04, 1.0550e-04, 8.8353e-05, 8.5830e-05, 9.3646e-05, 8.2364e-05, 9.8419e-05, 9.2501e-05], device='cuda:1') 2022-12-07 06:33:41,601 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:33:43,091 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:34:13,679 INFO [train.py:873] (1/4) Epoch 2, batch 800, loss[loss=0.2633, simple_loss=0.2453, pruned_loss=0.1406, over 14422.00 frames. ], tot_loss[loss=0.255, simple_loss=0.2402, pruned_loss=0.1349, over 1943465.74 frames. ], batch size: 51, lr: 3.51e-02, grad_scale: 16.0 2022-12-07 06:34:18,510 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 06:34:45,243 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:34:53,071 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 2.975e+02 4.226e+02 5.564e+02 1.451e+03, threshold=8.453e+02, percent-clipped=7.0 2022-12-07 06:35:26,074 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:35:36,626 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:35:40,665 INFO [train.py:873] (1/4) Epoch 2, batch 900, loss[loss=0.2432, simple_loss=0.2444, pruned_loss=0.121, over 14295.00 frames. ], tot_loss[loss=0.2555, simple_loss=0.2409, pruned_loss=0.1351, over 1963748.22 frames. ], batch size: 28, lr: 3.50e-02, grad_scale: 16.0 2022-12-07 06:36:17,483 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:36:17,650 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3371, 0.7964, 0.7875, 0.6607, 0.7467, 0.8651, 1.1452, 0.6929], device='cuda:1'), covar=tensor([0.0762, 0.2438, 0.0737, 0.1547, 0.0781, 0.0472, 0.0746, 0.0868], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0025, 0.0025, 0.0022, 0.0021, 0.0024, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([3.6196e-05, 4.4920e-05, 4.2371e-05, 4.1053e-05, 3.4275e-05, 4.0621e-05, 3.6069e-05, 3.4124e-05], device='cuda:1') 2022-12-07 06:36:19,007 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 3.358e+02 4.277e+02 5.886e+02 1.045e+03, threshold=8.554e+02, percent-clipped=3.0 2022-12-07 06:36:24,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6620, 3.4157, 3.4224, 3.7615, 3.7695, 3.6948, 3.7830, 3.3326], device='cuda:1'), covar=tensor([0.0391, 0.0809, 0.0440, 0.0407, 0.0454, 0.0477, 0.0579, 0.0562], device='cuda:1'), in_proj_covar=tensor([0.0060, 0.0109, 0.0075, 0.0070, 0.0078, 0.0082, 0.0104, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 06:36:59,512 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-07 06:37:06,427 INFO [train.py:873] (1/4) Epoch 2, batch 1000, loss[loss=0.267, simple_loss=0.2382, pruned_loss=0.1479, over 8673.00 frames. ], tot_loss[loss=0.2545, simple_loss=0.2402, pruned_loss=0.1344, over 1962776.39 frames. ], batch size: 100, lr: 3.48e-02, grad_scale: 16.0 2022-12-07 06:37:46,391 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 3.331e+02 4.272e+02 5.638e+02 1.026e+03, threshold=8.545e+02, percent-clipped=1.0 2022-12-07 06:37:57,595 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:38:13,587 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 06:38:34,124 INFO [train.py:873] (1/4) Epoch 2, batch 1100, loss[loss=0.2609, simple_loss=0.2486, pruned_loss=0.1366, over 14307.00 frames. ], tot_loss[loss=0.2556, simple_loss=0.2408, pruned_loss=0.1352, over 1930500.58 frames. ], batch size: 46, lr: 3.47e-02, grad_scale: 8.0 2022-12-07 06:38:41,123 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-07 06:39:13,894 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 3.265e+02 4.282e+02 5.954e+02 1.002e+03, threshold=8.564e+02, percent-clipped=5.0 2022-12-07 06:39:58,104 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:39:59,798 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0024, 3.7479, 4.1101, 3.5946, 3.8787, 3.9452, 1.8629, 3.9595], device='cuda:1'), covar=tensor([0.0186, 0.0258, 0.0325, 0.0260, 0.0236, 0.0162, 0.2730, 0.0171], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0076, 0.0081, 0.0065, 0.0106, 0.0068, 0.0121, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 06:40:00,478 INFO [train.py:873] (1/4) Epoch 2, batch 1200, loss[loss=0.2397, simple_loss=0.2364, pruned_loss=0.1215, over 14294.00 frames. ], tot_loss[loss=0.2539, simple_loss=0.2398, pruned_loss=0.134, over 1986503.22 frames. ], batch size: 63, lr: 3.45e-02, grad_scale: 8.0 2022-12-07 06:40:39,306 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 06:40:40,234 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.965e+02 3.180e+02 4.401e+02 5.751e+02 1.287e+03, threshold=8.803e+02, percent-clipped=4.0 2022-12-07 06:40:51,383 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:41:08,590 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 06:41:27,154 INFO [train.py:873] (1/4) Epoch 2, batch 1300, loss[loss=0.2015, simple_loss=0.1773, pruned_loss=0.1128, over 1168.00 frames. ], tot_loss[loss=0.2537, simple_loss=0.2393, pruned_loss=0.1341, over 1969302.58 frames. ], batch size: 100, lr: 3.44e-02, grad_scale: 8.0 2022-12-07 06:41:29,931 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8503, 1.7142, 2.9838, 2.3000, 3.0144, 1.7279, 2.6569, 2.7410], device='cuda:1'), covar=tensor([0.0314, 0.3416, 0.0210, 0.5047, 0.0171, 0.2955, 0.0643, 0.0211], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0277, 0.0141, 0.0378, 0.0116, 0.0293, 0.0191, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0003, 0.0001, 0.0003, 0.0002, 0.0001], device='cuda:1') 2022-12-07 06:41:33,311 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2230, 0.7504, 1.4662, 0.8779, 1.2279, 1.2636, 0.7441, 0.9655], device='cuda:1'), covar=tensor([0.0316, 0.0645, 0.0202, 0.0506, 0.0265, 0.0250, 0.0632, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0051, 0.0051, 0.0049, 0.0047, 0.0051, 0.0048, 0.0091, 0.0055], device='cuda:1'), out_proj_covar=tensor([6.9142e-05, 6.9747e-05, 6.5624e-05, 5.9390e-05, 6.3199e-05, 6.2428e-05, 1.2022e-04, 7.4024e-05], device='cuda:1') 2022-12-07 06:41:41,901 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.22 vs. limit=2.0 2022-12-07 06:41:47,168 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 06:41:49,582 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:42:05,268 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6643, 0.6441, 0.2588, 0.5268, 0.5987, 0.4996, 0.7173, 0.5483], device='cuda:1'), covar=tensor([0.0095, 0.0156, 0.0065, 0.0058, 0.0046, 0.0104, 0.0044, 0.0093], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0025, 0.0026, 0.0022, 0.0021, 0.0025, 0.0019, 0.0020], device='cuda:1'), out_proj_covar=tensor([3.9758e-05, 4.6458e-05, 4.4183e-05, 4.3244e-05, 3.5596e-05, 4.4816e-05, 3.6437e-05, 3.5392e-05], device='cuda:1') 2022-12-07 06:42:06,752 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.973e+02 3.684e+02 4.915e+02 1.042e+03, threshold=7.368e+02, percent-clipped=2.0 2022-12-07 06:42:08,142 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 06:42:17,212 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:42:22,319 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:42:24,215 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.97 vs. limit=2.0 2022-12-07 06:42:53,350 INFO [train.py:873] (1/4) Epoch 2, batch 1400, loss[loss=0.2564, simple_loss=0.2435, pruned_loss=0.1346, over 14227.00 frames. ], tot_loss[loss=0.2508, simple_loss=0.238, pruned_loss=0.1318, over 1952715.17 frames. ], batch size: 94, lr: 3.42e-02, grad_scale: 8.0 2022-12-07 06:42:58,578 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:42:59,894 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 06:43:01,137 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2063, 0.9324, 1.2121, 1.0475, 1.1144, 1.2291, 0.8904, 1.0434], device='cuda:1'), covar=tensor([0.2252, 0.1287, 0.0789, 0.0391, 0.0523, 0.0438, 0.0900, 0.0491], device='cuda:1'), in_proj_covar=tensor([0.0111, 0.0051, 0.0039, 0.0041, 0.0052, 0.0039, 0.0050, 0.0053], device='cuda:1'), out_proj_covar=tensor([2.1398e-04, 1.0871e-04, 9.2031e-05, 8.8546e-05, 1.0149e-04, 8.3245e-05, 1.0664e-04, 1.0390e-04], device='cuda:1') 2022-12-07 06:43:13,805 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:15,610 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:20,627 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8042, 1.8708, 1.5830, 1.9648, 2.0032, 1.9798, 1.6222, 1.5012], device='cuda:1'), covar=tensor([0.0163, 0.0286, 0.0640, 0.0097, 0.0133, 0.0102, 0.0308, 0.0524], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0157, 0.0260, 0.0120, 0.0115, 0.0118, 0.0153, 0.0293], device='cuda:1'), out_proj_covar=tensor([9.0767e-05, 1.0830e-04, 1.7040e-04, 7.9944e-05, 8.4077e-05, 8.4707e-05, 1.0669e-04, 1.8944e-04], device='cuda:1') 2022-12-07 06:43:32,036 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.31 vs. limit=2.0 2022-12-07 06:43:33,021 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 3.417e+02 4.711e+02 5.630e+02 1.118e+03, threshold=9.423e+02, percent-clipped=7.0 2022-12-07 06:43:36,693 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:44:06,574 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:44:20,924 INFO [train.py:873] (1/4) Epoch 2, batch 1500, loss[loss=0.2322, simple_loss=0.1881, pruned_loss=0.1381, over 1223.00 frames. ], tot_loss[loss=0.2505, simple_loss=0.2378, pruned_loss=0.1316, over 1971630.23 frames. ], batch size: 100, lr: 3.41e-02, grad_scale: 8.0 2022-12-07 06:44:30,445 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:44:59,853 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 3.213e+02 4.122e+02 5.427e+02 9.769e+02, threshold=8.244e+02, percent-clipped=1.0 2022-12-07 06:45:05,735 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:45:46,188 INFO [train.py:873] (1/4) Epoch 2, batch 1600, loss[loss=0.2538, simple_loss=0.2404, pruned_loss=0.1336, over 14284.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.2374, pruned_loss=0.1316, over 1943211.59 frames. ], batch size: 39, lr: 3.39e-02, grad_scale: 8.0 2022-12-07 06:46:04,018 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2087, 3.5408, 2.5222, 3.6949, 3.1177, 3.5756, 3.4294, 2.3478], device='cuda:1'), covar=tensor([0.0298, 0.0441, 0.2546, 0.0195, 0.0323, 0.0163, 0.0460, 0.3604], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0171, 0.0279, 0.0129, 0.0123, 0.0126, 0.0169, 0.0309], device='cuda:1'), out_proj_covar=tensor([9.6671e-05, 1.1901e-04, 1.8256e-04, 8.6027e-05, 9.1708e-05, 9.0667e-05, 1.1930e-04, 2.0003e-04], device='cuda:1') 2022-12-07 06:46:26,065 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 3.267e+02 4.651e+02 6.782e+02 1.677e+03, threshold=9.302e+02, percent-clipped=10.0 2022-12-07 06:46:35,486 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.54 vs. limit=2.0 2022-12-07 06:46:55,190 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8769, 3.4500, 3.9770, 3.2654, 3.7386, 3.7995, 1.7420, 3.6912], device='cuda:1'), covar=tensor([0.0212, 0.0391, 0.0387, 0.0395, 0.0309, 0.0236, 0.3123, 0.0235], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0076, 0.0080, 0.0063, 0.0105, 0.0071, 0.0118, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 06:47:02,748 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0286, 1.5280, 3.7532, 2.0130, 3.7961, 3.5092, 2.7301, 4.1401], device='cuda:1'), covar=tensor([0.0153, 0.2888, 0.0332, 0.2266, 0.0248, 0.0333, 0.0586, 0.0164], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0116, 0.0070, 0.0126, 0.0086, 0.0075, 0.0068, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 06:47:13,308 INFO [train.py:873] (1/4) Epoch 2, batch 1700, loss[loss=0.2101, simple_loss=0.2124, pruned_loss=0.1039, over 13664.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.2376, pruned_loss=0.1315, over 1920513.52 frames. ], batch size: 17, lr: 3.38e-02, grad_scale: 8.0 2022-12-07 06:47:30,392 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:47:53,364 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.995e+02 4.223e+02 5.322e+02 8.026e+02, threshold=8.446e+02, percent-clipped=0.0 2022-12-07 06:48:02,364 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.19 vs. limit=2.0 2022-12-07 06:48:21,610 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:48:39,297 INFO [train.py:873] (1/4) Epoch 2, batch 1800, loss[loss=0.276, simple_loss=0.2559, pruned_loss=0.148, over 14221.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.238, pruned_loss=0.1313, over 1942953.53 frames. ], batch size: 35, lr: 3.37e-02, grad_scale: 8.0 2022-12-07 06:48:42,131 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6396, 5.0020, 5.0903, 5.6703, 5.5232, 4.2570, 5.8358, 5.6735], device='cuda:1'), covar=tensor([0.0421, 0.0537, 0.0556, 0.0620, 0.0428, 0.0402, 0.0317, 0.0594], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0066, 0.0091, 0.0085, 0.0087, 0.0057, 0.0077, 0.0085], device='cuda:1'), out_proj_covar=tensor([1.3216e-04, 1.1152e-04, 1.4552e-04, 1.3715e-04, 1.4061e-04, 9.1743e-05, 1.2892e-04, 1.3788e-04], device='cuda:1') 2022-12-07 06:48:44,059 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5896, 1.9480, 3.7779, 2.6494, 3.4378, 1.8875, 3.2645, 3.5152], device='cuda:1'), covar=tensor([0.0212, 0.4451, 0.0225, 0.7876, 0.0152, 0.3741, 0.0478, 0.0144], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0273, 0.0146, 0.0389, 0.0119, 0.0305, 0.0199, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0003, 0.0001, 0.0003, 0.0002, 0.0001], device='cuda:1') 2022-12-07 06:48:44,775 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:48:59,794 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.88 vs. limit=5.0 2022-12-07 06:49:04,571 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0467, 4.4109, 4.5129, 4.9635, 4.7470, 4.5415, 4.9816, 4.5543], device='cuda:1'), covar=tensor([0.0321, 0.1093, 0.0305, 0.0386, 0.0572, 0.0382, 0.0563, 0.0427], device='cuda:1'), in_proj_covar=tensor([0.0067, 0.0120, 0.0076, 0.0078, 0.0084, 0.0085, 0.0115, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 06:49:19,720 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.949e+02 3.801e+02 5.110e+02 1.024e+03, threshold=7.602e+02, percent-clipped=3.0 2022-12-07 06:49:24,157 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:49:25,775 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:50:07,242 INFO [train.py:873] (1/4) Epoch 2, batch 1900, loss[loss=0.2158, simple_loss=0.2259, pruned_loss=0.1029, over 14036.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.237, pruned_loss=0.1301, over 1984907.65 frames. ], batch size: 26, lr: 3.35e-02, grad_scale: 8.0 2022-12-07 06:50:08,451 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:50:17,873 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:50:46,648 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.987e+02 3.780e+02 5.318e+02 1.239e+03, threshold=7.560e+02, percent-clipped=8.0 2022-12-07 06:51:33,030 INFO [train.py:873] (1/4) Epoch 2, batch 2000, loss[loss=0.2183, simple_loss=0.2231, pruned_loss=0.1068, over 13935.00 frames. ], tot_loss[loss=0.249, simple_loss=0.2374, pruned_loss=0.1303, over 1983022.91 frames. ], batch size: 20, lr: 3.34e-02, grad_scale: 8.0 2022-12-07 06:51:50,790 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:51:51,712 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:12,819 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 3.149e+02 4.158e+02 5.178e+02 1.006e+03, threshold=8.316e+02, percent-clipped=10.0 2022-12-07 06:52:32,382 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:41,414 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:44,055 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:54,899 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:59,449 INFO [train.py:873] (1/4) Epoch 2, batch 2100, loss[loss=0.2674, simple_loss=0.237, pruned_loss=0.1489, over 3827.00 frames. ], tot_loss[loss=0.2489, simple_loss=0.2374, pruned_loss=0.1302, over 2032116.57 frames. ], batch size: 100, lr: 3.32e-02, grad_scale: 8.0 2022-12-07 06:53:05,110 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:14,468 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1611, 3.3521, 4.2773, 3.8167, 4.2915, 3.3908, 3.9025, 2.9364], device='cuda:1'), covar=tensor([0.0094, 0.0207, 0.0100, 0.0187, 0.0098, 0.0493, 0.0055, 0.0488], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0104, 0.0078, 0.0099, 0.0096, 0.0153, 0.0063, 0.0145], device='cuda:1'), out_proj_covar=tensor([9.9894e-05, 1.2521e-04, 9.7806e-05, 1.2189e-04, 1.1633e-04, 1.8202e-04, 7.2015e-05, 1.6550e-04], device='cuda:1') 2022-12-07 06:53:22,858 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:39,633 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 3.245e+02 4.428e+02 6.131e+02 1.111e+03, threshold=8.856e+02, percent-clipped=8.0 2022-12-07 06:53:46,126 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:48,052 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:51,592 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 06:54:09,303 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:54:26,278 INFO [train.py:873] (1/4) Epoch 2, batch 2200, loss[loss=0.2141, simple_loss=0.2098, pruned_loss=0.1092, over 5997.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.2367, pruned_loss=0.1296, over 2001555.76 frames. ], batch size: 100, lr: 3.31e-02, grad_scale: 8.0 2022-12-07 06:54:32,304 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:54:47,944 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7698, 1.4599, 1.3442, 1.0496, 1.4155, 1.3262, 1.5964, 1.5957], device='cuda:1'), covar=tensor([0.0245, 0.1823, 0.0513, 0.0957, 0.0532, 0.0495, 0.0597, 0.0282], device='cuda:1'), in_proj_covar=tensor([0.0046, 0.0119, 0.0056, 0.0084, 0.0054, 0.0054, 0.0051, 0.0047], device='cuda:1'), out_proj_covar=tensor([8.2680e-05, 1.9824e-04, 1.0138e-04, 1.4964e-04, 1.0546e-04, 1.0326e-04, 1.0238e-04, 8.7831e-05], device='cuda:1') 2022-12-07 06:54:53,189 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-07 06:55:01,579 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:55:02,045 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 06:55:05,584 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.644e+01 3.042e+02 4.179e+02 5.699e+02 1.039e+03, threshold=8.358e+02, percent-clipped=2.0 2022-12-07 06:55:53,425 INFO [train.py:873] (1/4) Epoch 2, batch 2300, loss[loss=0.2717, simple_loss=0.2389, pruned_loss=0.1523, over 6908.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.2353, pruned_loss=0.129, over 1937966.28 frames. ], batch size: 100, lr: 3.30e-02, grad_scale: 8.0 2022-12-07 06:56:17,930 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.05 vs. limit=5.0 2022-12-07 06:56:33,304 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 3.202e+02 3.964e+02 5.237e+02 8.345e+02, threshold=7.929e+02, percent-clipped=0.0 2022-12-07 06:56:33,715 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6421, 1.2448, 1.9933, 1.7090, 1.9446, 1.9414, 1.5891, 1.9891], device='cuda:1'), covar=tensor([0.0335, 0.0809, 0.0127, 0.0349, 0.0140, 0.0090, 0.0457, 0.0091], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0117, 0.0058, 0.0080, 0.0066, 0.0069, 0.0056, 0.0056], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 06:56:33,727 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2144, 0.9431, 1.4537, 1.2618, 1.2493, 1.1727, 0.8256, 1.1173], device='cuda:1'), covar=tensor([0.0259, 0.0386, 0.0114, 0.0424, 0.0275, 0.0269, 0.0498, 0.0306], device='cuda:1'), in_proj_covar=tensor([0.0055, 0.0055, 0.0053, 0.0057, 0.0053, 0.0050, 0.0095, 0.0060], device='cuda:1'), out_proj_covar=tensor([7.9181e-05, 7.6615e-05, 7.5421e-05, 7.6812e-05, 6.8671e-05, 7.1016e-05, 1.3161e-04, 8.4211e-05], device='cuda:1') 2022-12-07 06:57:00,551 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:57:20,362 INFO [train.py:873] (1/4) Epoch 2, batch 2400, loss[loss=0.1901, simple_loss=0.171, pruned_loss=0.1046, over 2647.00 frames. ], tot_loss[loss=0.247, simple_loss=0.2353, pruned_loss=0.1293, over 1905053.98 frames. ], batch size: 100, lr: 3.28e-02, grad_scale: 8.0 2022-12-07 06:57:22,909 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2071, 1.7688, 3.8875, 2.1274, 3.8702, 3.7609, 3.2617, 4.3571], device='cuda:1'), covar=tensor([0.0153, 0.2593, 0.0345, 0.2011, 0.0230, 0.0241, 0.0402, 0.0141], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0122, 0.0077, 0.0130, 0.0093, 0.0079, 0.0072, 0.0075], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 06:58:03,556 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 3.087e+02 4.289e+02 5.805e+02 1.264e+03, threshold=8.579e+02, percent-clipped=4.0 2022-12-07 06:58:07,916 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:58:30,116 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:58:50,342 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:58:50,994 INFO [train.py:873] (1/4) Epoch 2, batch 2500, loss[loss=0.2513, simple_loss=0.2365, pruned_loss=0.133, over 14256.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.2358, pruned_loss=0.1288, over 1933933.58 frames. ], batch size: 80, lr: 3.27e-02, grad_scale: 8.0 2022-12-07 06:58:57,298 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:21,257 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:22,234 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:59:30,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 3.039e+02 4.037e+02 4.970e+02 1.749e+03, threshold=8.074e+02, percent-clipped=6.0 2022-12-07 06:59:36,266 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2294, 0.9861, 1.2288, 0.5896, 0.7085, 0.7592, 0.9133, 0.9463], device='cuda:1'), covar=tensor([0.0097, 0.0412, 0.0190, 0.0225, 0.0341, 0.0391, 0.0178, 0.0150], device='cuda:1'), in_proj_covar=tensor([0.0045, 0.0117, 0.0055, 0.0084, 0.0051, 0.0052, 0.0049, 0.0047], device='cuda:1'), out_proj_covar=tensor([8.2557e-05, 1.9727e-04, 1.0143e-04, 1.5188e-04, 1.0524e-04, 1.0070e-04, 1.0087e-04, 9.0626e-05], device='cuda:1') 2022-12-07 06:59:37,900 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:42,103 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 06:59:55,395 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:00:16,956 INFO [train.py:873] (1/4) Epoch 2, batch 2600, loss[loss=0.1806, simple_loss=0.1999, pruned_loss=0.08066, over 14613.00 frames. ], tot_loss[loss=0.2448, simple_loss=0.2347, pruned_loss=0.1275, over 1982479.04 frames. ], batch size: 22, lr: 3.26e-02, grad_scale: 8.0 2022-12-07 07:00:26,156 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 07:00:56,780 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 3.080e+02 4.222e+02 5.388e+02 1.025e+03, threshold=8.444e+02, percent-clipped=5.0 2022-12-07 07:01:24,275 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:01:44,396 INFO [train.py:873] (1/4) Epoch 2, batch 2700, loss[loss=0.2419, simple_loss=0.2265, pruned_loss=0.1287, over 14399.00 frames. ], tot_loss[loss=0.2434, simple_loss=0.2337, pruned_loss=0.1265, over 1972938.18 frames. ], batch size: 41, lr: 3.24e-02, grad_scale: 8.0 2022-12-07 07:02:06,312 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:02:24,379 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 3.215e+02 3.934e+02 5.144e+02 9.049e+02, threshold=7.869e+02, percent-clipped=4.0 2022-12-07 07:02:28,960 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:09,543 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:10,327 INFO [train.py:873] (1/4) Epoch 2, batch 2800, loss[loss=0.2284, simple_loss=0.1989, pruned_loss=0.129, over 2640.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.234, pruned_loss=0.1262, over 1983038.01 frames. ], batch size: 100, lr: 3.23e-02, grad_scale: 8.0 2022-12-07 07:03:13,467 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 07:03:35,486 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5290, 4.8332, 4.7876, 5.4639, 4.9762, 4.7457, 5.4173, 4.7773], device='cuda:1'), covar=tensor([0.0280, 0.0916, 0.0275, 0.0330, 0.0777, 0.0351, 0.0512, 0.0450], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0134, 0.0083, 0.0084, 0.0092, 0.0091, 0.0126, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:03:38,595 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-07 07:03:38,847 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:03:42,139 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:46,544 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0655, 2.1080, 1.9729, 2.1006, 1.8164, 1.7721, 2.0735, 2.1415], device='cuda:1'), covar=tensor([0.0954, 0.0730, 0.0966, 0.0780, 0.0967, 0.0755, 0.0894, 0.0704], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0069, 0.0095, 0.0086, 0.0094, 0.0064, 0.0087, 0.0085], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:03:50,604 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 3.431e+02 4.642e+02 5.590e+02 1.030e+03, threshold=9.284e+02, percent-clipped=5.0 2022-12-07 07:03:50,764 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:55,111 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8834, 0.8003, 0.7353, 0.6985, 0.4480, 0.6604, 1.0418, 0.4977], device='cuda:1'), covar=tensor([0.0100, 0.0262, 0.0249, 0.0180, 0.1439, 0.0425, 0.0123, 0.0617], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0015, 0.0015, 0.0017, 0.0015, 0.0016, 0.0014, 0.0016], device='cuda:1'), out_proj_covar=tensor([2.3924e-05, 2.7082e-05, 2.8500e-05, 2.6896e-05, 2.7722e-05, 2.4185e-05, 2.9344e-05, 2.7480e-05], device='cuda:1') 2022-12-07 07:03:58,245 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:04:09,050 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7598, 1.4027, 2.0683, 1.7732, 2.0176, 1.6207, 1.7948, 1.8968], device='cuda:1'), covar=tensor([0.0117, 0.0380, 0.0029, 0.0104, 0.0045, 0.0109, 0.0057, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0263, 0.0100, 0.0192, 0.0125, 0.0155, 0.0138, 0.0296], device='cuda:1'), out_proj_covar=tensor([1.2351e-04, 2.2475e-04, 8.3123e-05, 1.6044e-04, 1.1408e-04, 1.3619e-04, 1.3926e-04, 2.4871e-04], device='cuda:1') 2022-12-07 07:04:23,704 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:04:29,856 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:04:29,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5343, 2.1424, 3.1142, 3.0659, 2.6601, 2.1659, 2.8257, 2.1532], device='cuda:1'), covar=tensor([0.0133, 0.0225, 0.0136, 0.0176, 0.0125, 0.0521, 0.0059, 0.0478], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0111, 0.0087, 0.0105, 0.0101, 0.0159, 0.0065, 0.0149], device='cuda:1'), out_proj_covar=tensor([1.0563e-04, 1.3769e-04, 1.1752e-04, 1.3294e-04, 1.2740e-04, 1.9741e-04, 7.8095e-05, 1.7677e-04], device='cuda:1') 2022-12-07 07:04:37,270 INFO [train.py:873] (1/4) Epoch 2, batch 2900, loss[loss=0.23, simple_loss=0.214, pruned_loss=0.123, over 5987.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.234, pruned_loss=0.1262, over 1967788.51 frames. ], batch size: 100, lr: 3.22e-02, grad_scale: 8.0 2022-12-07 07:04:43,765 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:05:18,328 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.981e+02 3.992e+02 4.851e+02 1.210e+03, threshold=7.983e+02, percent-clipped=1.0 2022-12-07 07:05:23,803 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:06:04,702 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 07:06:05,154 INFO [train.py:873] (1/4) Epoch 2, batch 3000, loss[loss=0.2381, simple_loss=0.2334, pruned_loss=0.1214, over 14183.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.2325, pruned_loss=0.1249, over 1959610.25 frames. ], batch size: 89, lr: 3.21e-02, grad_scale: 8.0 2022-12-07 07:06:05,155 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 07:06:13,212 INFO [train.py:905] (1/4) Epoch 2, validation: loss=0.1433, simple_loss=0.1828, pruned_loss=0.05186, over 857387.00 frames. 2022-12-07 07:06:13,212 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 07:06:25,142 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 07:06:29,212 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0476, 0.7644, 1.1917, 0.7001, 0.7267, 0.8028, 0.8534, 0.7838], device='cuda:1'), covar=tensor([0.0160, 0.0413, 0.0211, 0.0398, 0.0284, 0.0593, 0.0207, 0.0238], device='cuda:1'), in_proj_covar=tensor([0.0050, 0.0129, 0.0063, 0.0090, 0.0054, 0.0056, 0.0051, 0.0051], device='cuda:1'), out_proj_covar=tensor([9.7033e-05, 2.2164e-04, 1.1738e-04, 1.6646e-04, 1.1293e-04, 1.0984e-04, 1.0949e-04, 1.0196e-04], device='cuda:1') 2022-12-07 07:06:38,033 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4753, 3.1263, 3.1894, 3.2014, 3.3486, 3.2466, 3.3556, 3.4277], device='cuda:1'), covar=tensor([0.0703, 0.0799, 0.0777, 0.1059, 0.0433, 0.0534, 0.1124, 0.0834], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0113, 0.0142, 0.0131, 0.0116, 0.0128, 0.0146, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:06:51,449 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1165, 3.7908, 3.7572, 4.1348, 3.9195, 3.0963, 4.1860, 4.1974], device='cuda:1'), covar=tensor([0.0553, 0.0553, 0.0585, 0.0710, 0.0608, 0.0605, 0.0660, 0.0617], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0069, 0.0093, 0.0090, 0.0095, 0.0064, 0.0087, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:06:54,390 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 3.162e+02 4.298e+02 5.900e+02 1.176e+03, threshold=8.596e+02, percent-clipped=5.0 2022-12-07 07:06:55,658 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2022-12-07 07:06:56,323 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2022-12-07 07:07:20,608 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8107, 0.6745, 0.5601, 0.3140, 0.3429, 0.8136, 0.3545, 0.5939], device='cuda:1'), covar=tensor([0.0068, 0.0122, 0.0075, 0.0082, 0.0607, 0.0067, 0.0095, 0.0186], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0016, 0.0017, 0.0016, 0.0017, 0.0015, 0.0018], device='cuda:1'), out_proj_covar=tensor([2.7078e-05, 2.7469e-05, 3.1289e-05, 2.6784e-05, 3.0190e-05, 2.6885e-05, 3.2938e-05, 3.0921e-05], device='cuda:1') 2022-12-07 07:07:41,330 INFO [train.py:873] (1/4) Epoch 2, batch 3100, loss[loss=0.2741, simple_loss=0.2448, pruned_loss=0.1517, over 4931.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.2323, pruned_loss=0.1258, over 1910588.34 frames. ], batch size: 100, lr: 3.19e-02, grad_scale: 8.0 2022-12-07 07:08:09,615 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:08:22,248 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 3.394e+02 4.427e+02 5.401e+02 1.921e+03, threshold=8.855e+02, percent-clipped=7.0 2022-12-07 07:08:29,204 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:08:37,215 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3230, 1.8343, 3.8619, 2.0357, 3.9953, 4.1502, 3.6226, 4.7014], device='cuda:1'), covar=tensor([0.0181, 0.2553, 0.0326, 0.2139, 0.0236, 0.0196, 0.0329, 0.0118], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0127, 0.0078, 0.0134, 0.0097, 0.0085, 0.0076, 0.0077], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:08:51,482 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:08:52,450 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:08,154 INFO [train.py:873] (1/4) Epoch 2, batch 3200, loss[loss=0.2045, simple_loss=0.1706, pruned_loss=0.1193, over 1283.00 frames. ], tot_loss[loss=0.2415, simple_loss=0.2322, pruned_loss=0.1254, over 1948409.88 frames. ], batch size: 100, lr: 3.18e-02, grad_scale: 8.0 2022-12-07 07:09:10,141 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:10,988 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:09:15,754 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9042, 1.5419, 2.8676, 1.6304, 2.8249, 2.8851, 2.0626, 2.9939], device='cuda:1'), covar=tensor([0.0179, 0.1990, 0.0257, 0.1617, 0.0262, 0.0293, 0.0719, 0.0195], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0128, 0.0079, 0.0134, 0.0098, 0.0085, 0.0077, 0.0078], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:09:16,604 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1211, 1.0156, 1.3868, 1.0107, 1.0795, 1.1361, 0.7781, 0.9972], device='cuda:1'), covar=tensor([0.0459, 0.0655, 0.0178, 0.0601, 0.0356, 0.0320, 0.0863, 0.0438], device='cuda:1'), in_proj_covar=tensor([0.0050, 0.0055, 0.0051, 0.0057, 0.0052, 0.0053, 0.0099, 0.0057], device='cuda:1'), out_proj_covar=tensor([7.7799e-05, 8.2804e-05, 7.7542e-05, 8.3492e-05, 7.2857e-05, 7.9569e-05, 1.3992e-04, 8.6742e-05], device='cuda:1') 2022-12-07 07:09:18,356 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:24,220 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8502, 0.8205, 1.0036, 0.9269, 0.7574, 0.9439, 0.8716, 0.6556], device='cuda:1'), covar=tensor([0.1058, 0.0569, 0.0285, 0.0460, 0.0588, 0.0194, 0.0735, 0.0714], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0055, 0.0044, 0.0046, 0.0059, 0.0044, 0.0052, 0.0061], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:09:24,517 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 07:09:36,998 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6664, 1.0581, 1.3580, 1.3239, 1.0559, 1.3914, 1.3348, 1.1357], device='cuda:1'), covar=tensor([0.1973, 0.0981, 0.0872, 0.1030, 0.0813, 0.0423, 0.0731, 0.0842], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0055, 0.0044, 0.0046, 0.0058, 0.0044, 0.0050, 0.0060], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:09:43,217 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9591, 4.2288, 5.0486, 4.0048, 4.8962, 5.0571, 2.4198, 4.7322], device='cuda:1'), covar=tensor([0.0173, 0.0380, 0.0434, 0.0404, 0.0229, 0.0156, 0.2897, 0.0253], device='cuda:1'), in_proj_covar=tensor([0.0095, 0.0093, 0.0094, 0.0075, 0.0118, 0.0083, 0.0135, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:09:44,958 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:49,654 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.759e+01 3.135e+02 4.113e+02 5.521e+02 8.692e+02, threshold=8.227e+02, percent-clipped=0.0 2022-12-07 07:09:49,790 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:10:11,395 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:10:23,327 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4807, 4.7847, 5.0947, 5.3686, 5.0997, 4.5760, 5.4565, 4.7591], device='cuda:1'), covar=tensor([0.0247, 0.1103, 0.0230, 0.0422, 0.0599, 0.0417, 0.0456, 0.0448], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0137, 0.0084, 0.0081, 0.0091, 0.0090, 0.0126, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:10:25,193 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6994, 3.1581, 3.3550, 3.6125, 3.4879, 3.6339, 3.6430, 3.1782], device='cuda:1'), covar=tensor([0.0283, 0.1311, 0.0324, 0.0512, 0.0708, 0.0375, 0.0615, 0.0575], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0138, 0.0084, 0.0081, 0.0091, 0.0091, 0.0127, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:10:30,598 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1428, 0.9745, 1.0780, 0.6480, 0.7755, 0.8889, 1.0763, 0.6928], device='cuda:1'), covar=tensor([0.0144, 0.0543, 0.0315, 0.0283, 0.0456, 0.0360, 0.0295, 0.0307], device='cuda:1'), in_proj_covar=tensor([0.0052, 0.0132, 0.0062, 0.0090, 0.0056, 0.0058, 0.0052, 0.0053], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:10:35,549 INFO [train.py:873] (1/4) Epoch 2, batch 3300, loss[loss=0.2444, simple_loss=0.2369, pruned_loss=0.1259, over 14173.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.2319, pruned_loss=0.1254, over 1909383.31 frames. ], batch size: 89, lr: 3.17e-02, grad_scale: 8.0 2022-12-07 07:10:54,141 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:11:15,637 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.971e+02 4.122e+02 5.420e+02 1.001e+03, threshold=8.244e+02, percent-clipped=4.0 2022-12-07 07:11:27,246 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3651, 1.2784, 1.8061, 1.2486, 1.4169, 1.5331, 0.9298, 1.3975], device='cuda:1'), covar=tensor([0.0627, 0.0727, 0.0297, 0.0960, 0.0625, 0.0295, 0.1370, 0.0602], device='cuda:1'), in_proj_covar=tensor([0.0052, 0.0057, 0.0053, 0.0060, 0.0057, 0.0053, 0.0100, 0.0061], device='cuda:1'), out_proj_covar=tensor([8.1584e-05, 8.5988e-05, 8.0269e-05, 8.7352e-05, 8.0591e-05, 8.1372e-05, 1.4256e-04, 9.3369e-05], device='cuda:1') 2022-12-07 07:11:46,503 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:12:00,254 INFO [train.py:873] (1/4) Epoch 2, batch 3400, loss[loss=0.2374, simple_loss=0.2335, pruned_loss=0.1207, over 14192.00 frames. ], tot_loss[loss=0.2404, simple_loss=0.2312, pruned_loss=0.1248, over 1901654.30 frames. ], batch size: 84, lr: 3.16e-02, grad_scale: 8.0 2022-12-07 07:12:42,276 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.877e+02 3.778e+02 5.417e+02 1.188e+03, threshold=7.556e+02, percent-clipped=7.0 2022-12-07 07:12:45,990 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0176, 2.2525, 3.5039, 3.3873, 3.5084, 2.2894, 3.4950, 2.5540], device='cuda:1'), covar=tensor([0.0093, 0.0208, 0.0147, 0.0150, 0.0069, 0.0410, 0.0033, 0.0342], device='cuda:1'), in_proj_covar=tensor([0.0092, 0.0114, 0.0093, 0.0111, 0.0101, 0.0161, 0.0065, 0.0153], device='cuda:1'), out_proj_covar=tensor([1.1210e-04, 1.4418e-04, 1.2742e-04, 1.4517e-04, 1.2949e-04, 2.0732e-04, 8.1093e-05, 1.8544e-04], device='cuda:1') 2022-12-07 07:12:59,337 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8356, 2.3229, 3.2655, 2.0174, 2.1679, 2.1556, 0.9601, 2.3012], device='cuda:1'), covar=tensor([0.1058, 0.0608, 0.0453, 0.0799, 0.0439, 0.0650, 0.2066, 0.0967], device='cuda:1'), in_proj_covar=tensor([0.0050, 0.0057, 0.0052, 0.0058, 0.0055, 0.0051, 0.0097, 0.0059], device='cuda:1'), out_proj_covar=tensor([7.8483e-05, 8.5247e-05, 7.9019e-05, 8.5145e-05, 7.8961e-05, 7.8458e-05, 1.3964e-04, 9.0523e-05], device='cuda:1') 2022-12-07 07:13:08,646 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3474, 1.4679, 3.2953, 1.5481, 3.1207, 3.1626, 2.2166, 3.4774], device='cuda:1'), covar=tensor([0.0172, 0.2320, 0.0305, 0.2106, 0.0320, 0.0298, 0.0714, 0.0152], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0126, 0.0079, 0.0134, 0.0100, 0.0084, 0.0077, 0.0077], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:13:28,835 INFO [train.py:873] (1/4) Epoch 2, batch 3500, loss[loss=0.3043, simple_loss=0.2668, pruned_loss=0.1709, over 8609.00 frames. ], tot_loss[loss=0.2398, simple_loss=0.2311, pruned_loss=0.1243, over 1972373.93 frames. ], batch size: 100, lr: 3.15e-02, grad_scale: 8.0 2022-12-07 07:13:30,347 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:00,451 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:08,559 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 3.159e+02 4.112e+02 5.380e+02 8.623e+02, threshold=8.224e+02, percent-clipped=2.0 2022-12-07 07:14:08,686 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:14:11,088 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:26,052 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:40,140 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8351, 1.8922, 1.5732, 2.0705, 1.8669, 2.0864, 1.7520, 1.4883], device='cuda:1'), covar=tensor([0.0111, 0.0204, 0.0466, 0.0066, 0.0124, 0.0076, 0.0223, 0.0347], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0183, 0.0283, 0.0127, 0.0132, 0.0132, 0.0174, 0.0309], device='cuda:1'), out_proj_covar=tensor([1.0759e-04, 1.2995e-04, 1.9462e-04, 8.7192e-05, 9.8288e-05, 9.7265e-05, 1.2925e-04, 2.0755e-04], device='cuda:1') 2022-12-07 07:14:44,071 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.05 vs. limit=5.0 2022-12-07 07:14:50,085 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:53,273 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.16 vs. limit=5.0 2022-12-07 07:14:54,227 INFO [train.py:873] (1/4) Epoch 2, batch 3600, loss[loss=0.2246, simple_loss=0.1864, pruned_loss=0.1314, over 1228.00 frames. ], tot_loss[loss=0.2382, simple_loss=0.2304, pruned_loss=0.123, over 1914775.90 frames. ], batch size: 100, lr: 3.13e-02, grad_scale: 8.0 2022-12-07 07:15:04,183 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:15:20,943 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:15:22,793 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1290, 2.9113, 3.2156, 2.7856, 3.0653, 2.5912, 1.2567, 3.0460], device='cuda:1'), covar=tensor([0.0249, 0.0369, 0.0359, 0.0336, 0.0326, 0.0758, 0.2754, 0.0250], device='cuda:1'), in_proj_covar=tensor([0.0094, 0.0086, 0.0088, 0.0069, 0.0116, 0.0078, 0.0128, 0.0106], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:15:35,172 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.817e+02 3.568e+02 4.695e+02 1.140e+03, threshold=7.135e+02, percent-clipped=4.0 2022-12-07 07:15:43,762 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 07:15:56,934 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:16:02,121 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:16:21,368 INFO [train.py:873] (1/4) Epoch 2, batch 3700, loss[loss=0.2432, simple_loss=0.2307, pruned_loss=0.1278, over 10373.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.2312, pruned_loss=0.1235, over 2020409.84 frames. ], batch size: 100, lr: 3.12e-02, grad_scale: 8.0 2022-12-07 07:16:27,802 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.07 vs. limit=2.0 2022-12-07 07:16:31,672 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9486, 2.7534, 2.5937, 2.5731, 2.7887, 2.6459, 2.8050, 2.8817], device='cuda:1'), covar=tensor([0.0491, 0.0623, 0.0866, 0.0981, 0.0429, 0.0539, 0.0954, 0.0671], device='cuda:1'), in_proj_covar=tensor([0.0129, 0.0121, 0.0154, 0.0147, 0.0124, 0.0138, 0.0154, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:16:55,756 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 07:17:02,054 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 3.008e+02 3.803e+02 5.016e+02 1.141e+03, threshold=7.605e+02, percent-clipped=8.0 2022-12-07 07:17:15,783 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0163, 0.8147, 1.2730, 0.7301, 1.3681, 1.1918, 1.3220, 0.9271], device='cuda:1'), covar=tensor([0.1864, 0.1001, 0.0213, 0.1978, 0.0908, 0.0325, 0.0633, 0.0757], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0020, 0.0021, 0.0019, 0.0019, 0.0021, 0.0017, 0.0019], device='cuda:1'), out_proj_covar=tensor([4.2673e-05, 4.4891e-05, 4.2763e-05, 4.3899e-05, 3.9196e-05, 4.4945e-05, 3.7893e-05, 3.8344e-05], device='cuda:1') 2022-12-07 07:17:24,526 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3155, 1.9273, 2.5518, 2.6735, 2.0857, 1.9783, 2.6169, 1.9639], device='cuda:1'), covar=tensor([0.0104, 0.0178, 0.0171, 0.0120, 0.0131, 0.0357, 0.0043, 0.0289], device='cuda:1'), in_proj_covar=tensor([0.0096, 0.0124, 0.0104, 0.0116, 0.0106, 0.0167, 0.0070, 0.0159], device='cuda:1'), out_proj_covar=tensor([1.1828e-04, 1.5902e-04, 1.4385e-04, 1.5272e-04, 1.3908e-04, 2.2067e-04, 8.8993e-05, 1.9439e-04], device='cuda:1') 2022-12-07 07:17:29,849 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0854, 0.7817, 1.0344, 0.7248, 1.3154, 0.9065, 1.3218, 0.9466], device='cuda:1'), covar=tensor([0.0626, 0.0712, 0.0223, 0.0997, 0.0262, 0.0283, 0.0431, 0.0635], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0020, 0.0021, 0.0019, 0.0019, 0.0021, 0.0016, 0.0019], device='cuda:1'), out_proj_covar=tensor([4.2303e-05, 4.5273e-05, 4.2441e-05, 4.4115e-05, 3.8356e-05, 4.4603e-05, 3.7673e-05, 3.8499e-05], device='cuda:1') 2022-12-07 07:17:33,817 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 07:17:47,319 INFO [train.py:873] (1/4) Epoch 2, batch 3800, loss[loss=0.2198, simple_loss=0.1878, pruned_loss=0.1259, over 1230.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.231, pruned_loss=0.1242, over 1965131.63 frames. ], batch size: 100, lr: 3.11e-02, grad_scale: 8.0 2022-12-07 07:18:19,984 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:18:28,500 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 3.176e+02 4.502e+02 5.921e+02 1.352e+03, threshold=9.005e+02, percent-clipped=11.0 2022-12-07 07:18:46,219 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:02,215 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:10,355 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.86 vs. limit=5.0 2022-12-07 07:19:11,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 07:19:15,325 INFO [train.py:873] (1/4) Epoch 2, batch 3900, loss[loss=0.2679, simple_loss=0.2369, pruned_loss=0.1495, over 6042.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.2297, pruned_loss=0.1226, over 1958180.57 frames. ], batch size: 100, lr: 3.10e-02, grad_scale: 8.0 2022-12-07 07:19:27,735 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:29,582 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2525, 1.6214, 1.7183, 1.2396, 1.2670, 1.7850, 1.7952, 1.7016], device='cuda:1'), covar=tensor([0.3737, 0.1473, 0.2853, 0.3770, 0.0837, 0.1132, 0.0859, 0.0812], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0060, 0.0046, 0.0049, 0.0060, 0.0048, 0.0059, 0.0063], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:19:51,417 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6165, 1.9046, 4.1077, 2.0516, 4.3187, 4.4111, 3.9468, 4.7631], device='cuda:1'), covar=tensor([0.0193, 0.2285, 0.0313, 0.2212, 0.0215, 0.0210, 0.0214, 0.0185], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0121, 0.0077, 0.0133, 0.0096, 0.0082, 0.0075, 0.0077], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:19:55,407 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.890e+02 4.018e+02 5.176e+02 1.364e+03, threshold=8.037e+02, percent-clipped=4.0 2022-12-07 07:20:12,903 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:20:14,717 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:20:22,685 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:20:32,590 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0795, 2.4836, 3.4838, 3.3040, 2.9319, 2.3286, 2.8039, 2.4469], device='cuda:1'), covar=tensor([0.0064, 0.0140, 0.0111, 0.0125, 0.0080, 0.0312, 0.0030, 0.0258], device='cuda:1'), in_proj_covar=tensor([0.0094, 0.0120, 0.0102, 0.0116, 0.0106, 0.0161, 0.0071, 0.0156], device='cuda:1'), out_proj_covar=tensor([1.1711e-04, 1.5600e-04, 1.4084e-04, 1.5288e-04, 1.3926e-04, 2.1493e-04, 8.8662e-05, 1.9239e-04], device='cuda:1') 2022-12-07 07:20:40,966 INFO [train.py:873] (1/4) Epoch 2, batch 4000, loss[loss=0.252, simple_loss=0.2202, pruned_loss=0.1419, over 3883.00 frames. ], tot_loss[loss=0.2366, simple_loss=0.229, pruned_loss=0.1221, over 1949360.88 frames. ], batch size: 100, lr: 3.09e-02, grad_scale: 8.0 2022-12-07 07:21:03,668 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:21:06,616 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.98 vs. limit=2.0 2022-12-07 07:21:07,148 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:21:16,167 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 07:21:22,285 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 3.113e+02 4.062e+02 5.460e+02 1.002e+03, threshold=8.123e+02, percent-clipped=4.0 2022-12-07 07:21:44,107 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:08,542 INFO [train.py:873] (1/4) Epoch 2, batch 4100, loss[loss=0.2328, simple_loss=0.2224, pruned_loss=0.1216, over 13536.00 frames. ], tot_loss[loss=0.2368, simple_loss=0.2288, pruned_loss=0.1224, over 1886640.77 frames. ], batch size: 100, lr: 3.08e-02, grad_scale: 8.0 2022-12-07 07:22:26,535 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9055, 1.8600, 2.5417, 1.8450, 1.5064, 1.5671, 0.8936, 1.8007], device='cuda:1'), covar=tensor([0.0708, 0.0723, 0.0481, 0.0643, 0.0526, 0.1095, 0.2089, 0.0776], device='cuda:1'), in_proj_covar=tensor([0.0052, 0.0057, 0.0055, 0.0057, 0.0056, 0.0050, 0.0098, 0.0062], device='cuda:1'), out_proj_covar=tensor([8.5272e-05, 9.0619e-05, 8.4970e-05, 8.6428e-05, 8.2643e-05, 8.0407e-05, 1.4540e-04, 9.9462e-05], device='cuda:1') 2022-12-07 07:22:37,405 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:39,910 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:49,364 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.791e+01 3.003e+02 4.031e+02 5.512e+02 8.434e+02, threshold=8.062e+02, percent-clipped=2.0 2022-12-07 07:22:58,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2855, 3.9223, 3.8592, 3.8681, 4.0282, 3.9606, 4.1790, 4.1999], device='cuda:1'), covar=tensor([0.0384, 0.0575, 0.0686, 0.0921, 0.0349, 0.0357, 0.0557, 0.0615], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0126, 0.0162, 0.0165, 0.0127, 0.0145, 0.0161, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:23:05,164 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1275, 1.6193, 1.8420, 1.2507, 1.0970, 1.8679, 1.5601, 1.2644], device='cuda:1'), covar=tensor([0.3654, 0.1659, 0.1353, 0.2318, 0.0880, 0.0517, 0.1484, 0.1118], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0058, 0.0044, 0.0047, 0.0058, 0.0045, 0.0058, 0.0063], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:23:32,930 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:23:35,181 INFO [train.py:873] (1/4) Epoch 2, batch 4200, loss[loss=0.2809, simple_loss=0.2438, pruned_loss=0.159, over 4961.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.2289, pruned_loss=0.1228, over 1860963.21 frames. ], batch size: 100, lr: 3.07e-02, grad_scale: 8.0 2022-12-07 07:24:16,691 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 3.300e+02 4.114e+02 5.198e+02 1.405e+03, threshold=8.228e+02, percent-clipped=5.0 2022-12-07 07:24:22,331 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:24:34,022 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:24:36,590 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0986, 1.8016, 2.4405, 2.5991, 1.8230, 1.8411, 2.3238, 2.0854], device='cuda:1'), covar=tensor([0.0063, 0.0120, 0.0089, 0.0065, 0.0084, 0.0227, 0.0037, 0.0138], device='cuda:1'), in_proj_covar=tensor([0.0095, 0.0123, 0.0105, 0.0115, 0.0104, 0.0162, 0.0070, 0.0154], device='cuda:1'), out_proj_covar=tensor([1.2093e-04, 1.6158e-04, 1.4440e-04, 1.5282e-04, 1.3652e-04, 2.1864e-04, 8.9687e-05, 1.9014e-04], device='cuda:1') 2022-12-07 07:24:38,286 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5590, 2.0875, 2.7675, 2.5091, 2.6446, 2.4634, 2.7912, 2.1808], device='cuda:1'), covar=tensor([0.0161, 0.1217, 0.0056, 0.0485, 0.0133, 0.0291, 0.0252, 0.1037], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0283, 0.0116, 0.0206, 0.0141, 0.0168, 0.0161, 0.0316], device='cuda:1'), out_proj_covar=tensor([1.4150e-04, 2.5141e-04, 9.8649e-05, 1.7621e-04, 1.3221e-04, 1.5302e-04, 1.6355e-04, 2.7135e-04], device='cuda:1') 2022-12-07 07:24:47,753 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:01,559 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2022-12-07 07:25:02,802 INFO [train.py:873] (1/4) Epoch 2, batch 4300, loss[loss=0.2598, simple_loss=0.2407, pruned_loss=0.1395, over 10376.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.229, pruned_loss=0.1226, over 1894787.87 frames. ], batch size: 100, lr: 3.06e-02, grad_scale: 8.0 2022-12-07 07:25:11,975 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.31 vs. limit=2.0 2022-12-07 07:25:15,071 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:15,836 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:24,299 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:41,678 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:44,081 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 3.023e+02 3.927e+02 4.940e+02 9.062e+02, threshold=7.854e+02, percent-clipped=2.0 2022-12-07 07:25:52,622 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:57,494 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2022-12-07 07:26:17,403 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:29,330 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9364, 0.9744, 1.0977, 1.0208, 0.8074, 1.1200, 0.6848, 0.5762], device='cuda:1'), covar=tensor([0.1028, 0.0262, 0.0293, 0.0196, 0.0586, 0.0141, 0.0697, 0.0552], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0053, 0.0044, 0.0043, 0.0055, 0.0043, 0.0054, 0.0058], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:26:30,061 INFO [train.py:873] (1/4) Epoch 2, batch 4400, loss[loss=0.2092, simple_loss=0.1743, pruned_loss=0.122, over 1202.00 frames. ], tot_loss[loss=0.2379, simple_loss=0.2295, pruned_loss=0.1231, over 1861566.50 frames. ], batch size: 100, lr: 3.04e-02, grad_scale: 8.0 2022-12-07 07:26:46,416 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:48,586 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:54,435 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:01,429 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6821, 2.6052, 3.8927, 3.0629, 3.4741, 3.5287, 3.5974, 3.1150], device='cuda:1'), covar=tensor([0.0126, 0.1435, 0.0060, 0.0512, 0.0204, 0.0247, 0.0531, 0.1049], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0287, 0.0121, 0.0219, 0.0148, 0.0175, 0.0168, 0.0331], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 07:27:10,445 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:11,348 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.977e+02 4.090e+02 5.341e+02 9.805e+02, threshold=8.180e+02, percent-clipped=2.0 2022-12-07 07:27:26,566 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2817, 1.1667, 1.3247, 1.3557, 1.0228, 1.3156, 0.7569, 0.9919], device='cuda:1'), covar=tensor([0.2466, 0.0874, 0.0491, 0.0420, 0.0825, 0.0515, 0.1016, 0.0654], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0055, 0.0045, 0.0044, 0.0057, 0.0044, 0.0055, 0.0060], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:27:41,918 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:50,974 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:57,854 INFO [train.py:873] (1/4) Epoch 2, batch 4500, loss[loss=0.229, simple_loss=0.228, pruned_loss=0.115, over 14269.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.2286, pruned_loss=0.1214, over 1915732.09 frames. ], batch size: 46, lr: 3.03e-02, grad_scale: 8.0 2022-12-07 07:28:09,046 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:28:38,200 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 3.534e+02 4.689e+02 6.101e+02 2.207e+03, threshold=9.377e+02, percent-clipped=9.0 2022-12-07 07:28:46,088 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8942, 1.3189, 1.5484, 1.5709, 1.1416, 1.4982, 1.1599, 1.1146], device='cuda:1'), covar=tensor([0.2359, 0.0860, 0.0942, 0.0646, 0.1023, 0.0418, 0.1897, 0.1005], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0056, 0.0047, 0.0045, 0.0058, 0.0044, 0.0058, 0.0063], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 07:28:46,836 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3576, 4.7629, 4.8809, 5.5137, 5.1723, 3.9196, 5.5588, 5.5461], device='cuda:1'), covar=tensor([0.0688, 0.0695, 0.0756, 0.0515, 0.0598, 0.0573, 0.0584, 0.0536], device='cuda:1'), in_proj_covar=tensor([0.0093, 0.0073, 0.0093, 0.0090, 0.0099, 0.0065, 0.0085, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:29:01,384 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:07,357 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8020, 1.5735, 2.1195, 1.6627, 2.1091, 1.7275, 1.9299, 1.8572], device='cuda:1'), covar=tensor([0.0071, 0.0254, 0.0024, 0.0057, 0.0029, 0.0071, 0.0043, 0.0091], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0289, 0.0117, 0.0215, 0.0147, 0.0175, 0.0166, 0.0322], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 07:29:09,209 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5068, 2.0060, 4.6510, 3.1473, 4.6372, 2.2362, 3.8609, 4.5297], device='cuda:1'), covar=tensor([0.0138, 0.4129, 0.0170, 0.6377, 0.0073, 0.2679, 0.0452, 0.0064], device='cuda:1'), in_proj_covar=tensor([0.0196, 0.0295, 0.0148, 0.0405, 0.0134, 0.0306, 0.0232, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0004, 0.0001, 0.0003, 0.0002, 0.0001], device='cuda:1') 2022-12-07 07:29:23,549 INFO [train.py:873] (1/4) Epoch 2, batch 4600, loss[loss=0.2331, simple_loss=0.2375, pruned_loss=0.1143, over 14227.00 frames. ], tot_loss[loss=0.2373, simple_loss=0.2295, pruned_loss=0.1226, over 1974890.02 frames. ], batch size: 57, lr: 3.02e-02, grad_scale: 8.0 2022-12-07 07:29:27,099 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5429, 3.3731, 2.5665, 3.7092, 3.3261, 3.7188, 3.3599, 2.6230], device='cuda:1'), covar=tensor([0.0153, 0.0307, 0.2062, 0.0151, 0.0168, 0.0330, 0.0264, 0.2053], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0196, 0.0309, 0.0142, 0.0149, 0.0142, 0.0193, 0.0329], device='cuda:1'), out_proj_covar=tensor([1.1350e-04, 1.4055e-04, 2.1388e-04, 9.9396e-05, 1.1175e-04, 1.0849e-04, 1.4626e-04, 2.2254e-04], device='cuda:1') 2022-12-07 07:29:31,599 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:45,282 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:57,593 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:30:04,545 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.710e+02 3.814e+02 4.807e+02 1.336e+03, threshold=7.627e+02, percent-clipped=4.0 2022-12-07 07:30:08,004 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 07:30:27,152 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:30:28,385 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 07:30:42,043 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6372, 1.4978, 2.1074, 1.6567, 2.1318, 1.7820, 1.9080, 1.9077], device='cuda:1'), covar=tensor([0.0078, 0.0293, 0.0029, 0.0067, 0.0036, 0.0079, 0.0046, 0.0110], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0288, 0.0118, 0.0223, 0.0149, 0.0177, 0.0171, 0.0324], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 07:30:50,578 INFO [train.py:873] (1/4) Epoch 2, batch 4700, loss[loss=0.1979, simple_loss=0.1691, pruned_loss=0.1133, over 1299.00 frames. ], tot_loss[loss=0.2379, simple_loss=0.2302, pruned_loss=0.1228, over 1935041.15 frames. ], batch size: 100, lr: 3.01e-02, grad_scale: 8.0 2022-12-07 07:31:01,589 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:14,642 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:25,435 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:28,861 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9383, 3.6408, 3.6691, 3.6276, 3.6864, 3.8008, 3.9091, 3.9419], device='cuda:1'), covar=tensor([0.0705, 0.0441, 0.0535, 0.0988, 0.0362, 0.0383, 0.0795, 0.0612], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0129, 0.0161, 0.0180, 0.0135, 0.0152, 0.0169, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:31:30,462 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.849e+02 3.785e+02 5.259e+02 1.272e+03, threshold=7.569e+02, percent-clipped=8.0 2022-12-07 07:31:39,529 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 07:31:40,553 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0445, 2.0340, 1.8483, 2.0686, 1.8250, 1.9224, 2.0775, 2.1479], device='cuda:1'), covar=tensor([0.0980, 0.0882, 0.1021, 0.0826, 0.1200, 0.0709, 0.0875, 0.0845], device='cuda:1'), in_proj_covar=tensor([0.0091, 0.0073, 0.0091, 0.0089, 0.0097, 0.0064, 0.0085, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 07:31:55,988 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:56,824 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:09,919 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:16,668 INFO [train.py:873] (1/4) Epoch 2, batch 4800, loss[loss=0.2382, simple_loss=0.2288, pruned_loss=0.1238, over 4977.00 frames. ], tot_loss[loss=0.237, simple_loss=0.2295, pruned_loss=0.1223, over 1987305.32 frames. ], batch size: 100, lr: 3.00e-02, grad_scale: 8.0 2022-12-07 07:32:20,699 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0290, 1.9046, 1.9779, 2.0072, 2.0738, 1.8700, 1.0443, 1.8390], device='cuda:1'), covar=tensor([0.0227, 0.0253, 0.0428, 0.0153, 0.0203, 0.0347, 0.1363, 0.0250], device='cuda:1'), in_proj_covar=tensor([0.0094, 0.0096, 0.0096, 0.0071, 0.0122, 0.0082, 0.0133, 0.0111], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:32:21,247 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 07:32:51,094 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:57,419 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 3.011e+02 4.103e+02 5.225e+02 9.734e+02, threshold=8.206e+02, percent-clipped=3.0 2022-12-07 07:33:01,007 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6980, 2.2708, 3.0206, 3.0124, 2.4958, 2.2331, 2.8022, 2.3701], device='cuda:1'), covar=tensor([0.0055, 0.0116, 0.0110, 0.0086, 0.0067, 0.0218, 0.0035, 0.0173], device='cuda:1'), in_proj_covar=tensor([0.0104, 0.0128, 0.0120, 0.0123, 0.0110, 0.0172, 0.0078, 0.0159], device='cuda:1'), out_proj_covar=tensor([1.3331e-04, 1.7095e-04, 1.6581e-04, 1.6449e-04, 1.4702e-04, 2.3472e-04, 9.9341e-05, 1.9812e-04], device='cuda:1') 2022-12-07 07:33:03,950 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 07:33:04,125 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 07:33:06,032 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 07:33:15,507 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:33:27,416 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:33:28,476 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2022-12-07 07:33:42,230 INFO [train.py:873] (1/4) Epoch 2, batch 4900, loss[loss=0.2012, simple_loss=0.1883, pruned_loss=0.1071, over 3859.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.2295, pruned_loss=0.1223, over 1972624.96 frames. ], batch size: 100, lr: 2.99e-02, grad_scale: 8.0 2022-12-07 07:33:45,488 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3777, 1.7663, 1.2033, 1.0229, 1.2903, 1.3919, 0.8757, 1.2935], device='cuda:1'), covar=tensor([0.0354, 0.0294, 0.0785, 0.0250, 0.0266, 0.0288, 0.0189, 0.0209], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0013, 0.0014, 0.0013, 0.0016, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([2.4232e-05, 2.6162e-05, 2.8903e-05, 2.6722e-05, 2.7179e-05, 2.7949e-05, 3.1458e-05, 2.9506e-05], device='cuda:1') 2022-12-07 07:33:49,684 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:15,592 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:19,795 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:21,912 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 3.534e+02 4.435e+02 5.863e+02 9.969e+02, threshold=8.870e+02, percent-clipped=5.0 2022-12-07 07:34:30,740 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:44,296 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 07:34:56,798 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:08,167 INFO [train.py:873] (1/4) Epoch 2, batch 5000, loss[loss=0.2318, simple_loss=0.1947, pruned_loss=0.1344, over 1217.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.2293, pruned_loss=0.1219, over 1949893.33 frames. ], batch size: 100, lr: 2.98e-02, grad_scale: 8.0 2022-12-07 07:35:19,786 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:44,184 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:49,096 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 3.073e+02 4.044e+02 5.211e+02 8.552e+02, threshold=8.087e+02, percent-clipped=0.0 2022-12-07 07:36:01,963 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:03,093 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9115, 2.8119, 4.0523, 3.1605, 3.8560, 3.7050, 3.6215, 3.2497], device='cuda:1'), covar=tensor([0.0071, 0.0871, 0.0036, 0.0433, 0.0115, 0.0150, 0.0502, 0.0831], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0300, 0.0120, 0.0233, 0.0162, 0.0191, 0.0175, 0.0338], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 07:36:15,482 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:25,605 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:26,433 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9834, 0.5424, 1.0494, 0.6975, 1.5407, 0.8550, 1.3573, 0.8656], device='cuda:1'), covar=tensor([0.0662, 0.0637, 0.0287, 0.0666, 0.0570, 0.0242, 0.0378, 0.0265], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0020, 0.0018, 0.0017, 0.0022, 0.0017, 0.0018], device='cuda:1'), out_proj_covar=tensor([4.5748e-05, 4.4904e-05, 4.2676e-05, 4.4682e-05, 3.9088e-05, 4.7337e-05, 4.0803e-05, 3.9754e-05], device='cuda:1') 2022-12-07 07:36:35,311 INFO [train.py:873] (1/4) Epoch 2, batch 5100, loss[loss=0.224, simple_loss=0.2247, pruned_loss=0.1117, over 14259.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.2278, pruned_loss=0.1203, over 1968261.45 frames. ], batch size: 76, lr: 2.97e-02, grad_scale: 8.0 2022-12-07 07:36:56,656 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:37:16,599 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 3.113e+02 4.447e+02 5.356e+02 1.001e+03, threshold=8.894e+02, percent-clipped=4.0 2022-12-07 07:37:17,697 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2146, 1.5120, 1.3315, 1.7615, 1.2869, 1.2272, 1.3131, 1.6262], device='cuda:1'), covar=tensor([0.1970, 0.1126, 0.0424, 0.0282, 0.0791, 0.0321, 0.0345, 0.0675], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0015, 0.0013, 0.0015, 0.0014, 0.0016, 0.0015, 0.0016], device='cuda:1'), out_proj_covar=tensor([2.7787e-05, 2.8146e-05, 3.0283e-05, 2.7975e-05, 3.0181e-05, 2.9881e-05, 3.5896e-05, 3.1620e-05], device='cuda:1') 2022-12-07 07:37:34,948 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:37:39,632 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1129, 1.8647, 4.6057, 2.2003, 4.4156, 4.5796, 4.4437, 5.1325], device='cuda:1'), covar=tensor([0.0106, 0.2792, 0.0208, 0.2321, 0.0249, 0.0201, 0.0138, 0.0135], device='cuda:1'), in_proj_covar=tensor([0.0098, 0.0136, 0.0085, 0.0146, 0.0106, 0.0089, 0.0082, 0.0085], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:37:53,481 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 07:38:01,731 INFO [train.py:873] (1/4) Epoch 2, batch 5200, loss[loss=0.244, simple_loss=0.2298, pruned_loss=0.1291, over 14029.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.2278, pruned_loss=0.1205, over 1960144.54 frames. ], batch size: 29, lr: 2.96e-02, grad_scale: 8.0 2022-12-07 07:38:01,923 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:03,541 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7482, 2.6655, 2.4228, 2.3950, 2.7003, 2.5943, 2.7239, 2.7186], device='cuda:1'), covar=tensor([0.0612, 0.0779, 0.1230, 0.1938, 0.0471, 0.0551, 0.0821, 0.0657], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0138, 0.0170, 0.0199, 0.0135, 0.0163, 0.0171, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:38:16,631 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:38:16,882 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:35,737 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:40,758 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0216, 1.7118, 2.4118, 2.1153, 2.4054, 1.9084, 2.0606, 2.0882], device='cuda:1'), covar=tensor([0.0102, 0.0376, 0.0030, 0.0111, 0.0054, 0.0132, 0.0066, 0.0194], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0297, 0.0117, 0.0228, 0.0164, 0.0187, 0.0173, 0.0329], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 07:38:43,163 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 3.055e+02 3.810e+02 5.193e+02 9.032e+02, threshold=7.619e+02, percent-clipped=1.0 2022-12-07 07:38:47,204 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2022-12-07 07:38:54,892 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:39:28,088 INFO [train.py:873] (1/4) Epoch 2, batch 5300, loss[loss=0.2226, simple_loss=0.2215, pruned_loss=0.1119, over 14250.00 frames. ], tot_loss[loss=0.2351, simple_loss=0.2282, pruned_loss=0.121, over 1938099.67 frames. ], batch size: 39, lr: 2.95e-02, grad_scale: 8.0 2022-12-07 07:40:00,531 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 07:40:09,055 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.807e+01 2.959e+02 3.875e+02 4.793e+02 1.192e+03, threshold=7.749e+02, percent-clipped=2.0 2022-12-07 07:40:53,785 INFO [train.py:873] (1/4) Epoch 2, batch 5400, loss[loss=0.2129, simple_loss=0.2193, pruned_loss=0.1032, over 14282.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.2278, pruned_loss=0.1199, over 1944705.69 frames. ], batch size: 63, lr: 2.94e-02, grad_scale: 8.0 2022-12-07 07:41:35,174 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 3.363e+02 4.433e+02 5.944e+02 1.364e+03, threshold=8.866e+02, percent-clipped=4.0 2022-12-07 07:41:36,465 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-07 07:42:21,076 INFO [train.py:873] (1/4) Epoch 2, batch 5500, loss[loss=0.2344, simple_loss=0.1941, pruned_loss=0.1373, over 1234.00 frames. ], tot_loss[loss=0.2313, simple_loss=0.2259, pruned_loss=0.1184, over 1910162.44 frames. ], batch size: 100, lr: 2.93e-02, grad_scale: 8.0 2022-12-07 07:42:47,985 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.53 vs. limit=2.0 2022-12-07 07:42:54,607 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:43:01,929 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 3.268e+02 4.131e+02 4.754e+02 9.036e+02, threshold=8.263e+02, percent-clipped=2.0 2022-12-07 07:43:08,752 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:43:33,327 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2022-12-07 07:43:35,172 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:43:46,160 INFO [train.py:873] (1/4) Epoch 2, batch 5600, loss[loss=0.2447, simple_loss=0.2318, pruned_loss=0.1288, over 11164.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.2267, pruned_loss=0.1192, over 1947201.61 frames. ], batch size: 100, lr: 2.92e-02, grad_scale: 8.0 2022-12-07 07:44:28,274 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 3.284e+02 4.567e+02 6.071e+02 1.505e+03, threshold=9.133e+02, percent-clipped=13.0 2022-12-07 07:44:41,014 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3253, 1.6456, 3.8623, 1.9982, 3.9671, 3.8263, 3.2522, 4.4609], device='cuda:1'), covar=tensor([0.0145, 0.2490, 0.0329, 0.1877, 0.0287, 0.0326, 0.0425, 0.0122], device='cuda:1'), in_proj_covar=tensor([0.0099, 0.0136, 0.0088, 0.0143, 0.0109, 0.0093, 0.0083, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:45:08,159 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1286, 3.8625, 3.7714, 3.7292, 3.8933, 3.9367, 4.1844, 4.1198], device='cuda:1'), covar=tensor([0.0706, 0.0637, 0.1022, 0.1571, 0.0462, 0.0497, 0.0677, 0.0887], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0147, 0.0183, 0.0213, 0.0148, 0.0168, 0.0182, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:45:13,694 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8486, 1.0677, 1.1440, 1.1325, 0.9437, 1.1320, 0.9518, 0.8632], device='cuda:1'), covar=tensor([0.1913, 0.0955, 0.0329, 0.0365, 0.0709, 0.0364, 0.1168, 0.0946], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0064, 0.0050, 0.0053, 0.0064, 0.0047, 0.0061, 0.0078], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:45:14,446 INFO [train.py:873] (1/4) Epoch 2, batch 5700, loss[loss=0.2435, simple_loss=0.1963, pruned_loss=0.1454, over 1253.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.2268, pruned_loss=0.1191, over 1936207.61 frames. ], batch size: 100, lr: 2.91e-02, grad_scale: 8.0 2022-12-07 07:45:16,640 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.13 vs. limit=5.0 2022-12-07 07:45:55,408 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.840e+02 3.959e+02 4.964e+02 1.183e+03, threshold=7.919e+02, percent-clipped=6.0 2022-12-07 07:46:40,288 INFO [train.py:873] (1/4) Epoch 2, batch 5800, loss[loss=0.2472, simple_loss=0.2315, pruned_loss=0.1314, over 9503.00 frames. ], tot_loss[loss=0.2322, simple_loss=0.2265, pruned_loss=0.119, over 1956468.58 frames. ], batch size: 100, lr: 2.90e-02, grad_scale: 8.0 2022-12-07 07:46:56,623 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2051, 0.9275, 0.4284, 1.1871, 1.2853, 0.6958, 1.0174, 0.7032], device='cuda:1'), covar=tensor([0.0775, 0.0503, 0.0251, 0.0418, 0.0910, 0.0421, 0.0617, 0.1075], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0020, 0.0017, 0.0018, 0.0021, 0.0016, 0.0018], device='cuda:1'), out_proj_covar=tensor([4.5308e-05, 4.6276e-05, 4.4020e-05, 4.3040e-05, 4.3037e-05, 4.8181e-05, 4.1649e-05, 4.3755e-05], device='cuda:1') 2022-12-07 07:47:10,623 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3430, 1.2742, 0.7562, 0.9884, 1.5785, 0.8601, 1.2412, 1.0929], device='cuda:1'), covar=tensor([0.0991, 0.1351, 0.1521, 0.0520, 0.0239, 0.0295, 0.0182, 0.0561], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0014, 0.0016, 0.0014, 0.0017, 0.0014, 0.0016], device='cuda:1'), out_proj_covar=tensor([2.9517e-05, 2.8364e-05, 3.2097e-05, 3.1007e-05, 2.9476e-05, 3.1858e-05, 3.6092e-05, 3.4513e-05], device='cuda:1') 2022-12-07 07:47:21,674 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 3.070e+02 4.104e+02 5.303e+02 9.108e+02, threshold=8.208e+02, percent-clipped=3.0 2022-12-07 07:47:29,308 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:48:07,860 INFO [train.py:873] (1/4) Epoch 2, batch 5900, loss[loss=0.2677, simple_loss=0.2452, pruned_loss=0.1451, over 11231.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.2257, pruned_loss=0.1179, over 1964043.06 frames. ], batch size: 100, lr: 2.89e-02, grad_scale: 8.0 2022-12-07 07:48:11,325 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:48:28,200 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.82 vs. limit=2.0 2022-12-07 07:48:35,356 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1182, 3.3645, 3.8619, 4.0058, 3.8602, 3.7802, 3.8425, 3.5117], device='cuda:1'), covar=tensor([0.0759, 0.1737, 0.0519, 0.0747, 0.0896, 0.0760, 0.1175, 0.0914], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0151, 0.0097, 0.0092, 0.0096, 0.0092, 0.0143, 0.0112], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:48:49,278 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 3.527e+02 4.364e+02 5.819e+02 1.292e+03, threshold=8.728e+02, percent-clipped=7.0 2022-12-07 07:49:20,252 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.03 vs. limit=5.0 2022-12-07 07:49:33,859 INFO [train.py:873] (1/4) Epoch 2, batch 6000, loss[loss=0.2417, simple_loss=0.2444, pruned_loss=0.1195, over 14677.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.2261, pruned_loss=0.1176, over 2038869.33 frames. ], batch size: 33, lr: 2.88e-02, grad_scale: 8.0 2022-12-07 07:49:33,859 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 07:49:42,176 INFO [train.py:905] (1/4) Epoch 2, validation: loss=0.139, simple_loss=0.1792, pruned_loss=0.04941, over 857387.00 frames. 2022-12-07 07:49:42,176 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 07:50:24,252 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 3.403e+02 4.162e+02 5.697e+02 1.072e+03, threshold=8.324e+02, percent-clipped=3.0 2022-12-07 07:50:36,642 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=12.20 vs. limit=5.0 2022-12-07 07:50:41,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0031, 1.1629, 1.0275, 1.0302, 0.7129, 0.6826, 0.3929, 0.9693], device='cuda:1'), covar=tensor([0.0044, 0.0141, 0.0131, 0.0160, 0.0507, 0.0103, 0.0049, 0.0148], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0013, 0.0014, 0.0014, 0.0017, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([2.6478e-05, 2.7225e-05, 3.1830e-05, 2.9885e-05, 3.0241e-05, 3.2925e-05, 3.4855e-05, 3.3304e-05], device='cuda:1') 2022-12-07 07:51:09,282 INFO [train.py:873] (1/4) Epoch 2, batch 6100, loss[loss=0.2368, simple_loss=0.2246, pruned_loss=0.1244, over 14241.00 frames. ], tot_loss[loss=0.2307, simple_loss=0.2257, pruned_loss=0.1178, over 1982210.86 frames. ], batch size: 94, lr: 2.87e-02, grad_scale: 8.0 2022-12-07 07:51:20,946 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:51:50,708 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 3.186e+02 4.052e+02 4.953e+02 1.165e+03, threshold=8.105e+02, percent-clipped=5.0 2022-12-07 07:52:03,223 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 07:52:13,613 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:52:35,526 INFO [train.py:873] (1/4) Epoch 2, batch 6200, loss[loss=0.2501, simple_loss=0.2366, pruned_loss=0.1318, over 13537.00 frames. ], tot_loss[loss=0.2316, simple_loss=0.2265, pruned_loss=0.1183, over 2011972.52 frames. ], batch size: 100, lr: 2.86e-02, grad_scale: 8.0 2022-12-07 07:52:49,662 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8939, 1.6124, 3.0582, 1.4176, 2.9626, 2.9347, 1.9587, 3.0934], device='cuda:1'), covar=tensor([0.0183, 0.1948, 0.0204, 0.2008, 0.0219, 0.0264, 0.0649, 0.0159], device='cuda:1'), in_proj_covar=tensor([0.0098, 0.0133, 0.0086, 0.0140, 0.0107, 0.0091, 0.0079, 0.0084], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:53:05,548 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:53:08,932 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-07 07:53:17,360 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.778e+02 3.775e+02 5.188e+02 1.088e+03, threshold=7.550e+02, percent-clipped=5.0 2022-12-07 07:53:33,040 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.59 vs. limit=5.0 2022-12-07 07:54:02,989 INFO [train.py:873] (1/4) Epoch 2, batch 6300, loss[loss=0.2334, simple_loss=0.2298, pruned_loss=0.1186, over 14269.00 frames. ], tot_loss[loss=0.2311, simple_loss=0.2263, pruned_loss=0.118, over 2006975.85 frames. ], batch size: 76, lr: 2.86e-02, grad_scale: 8.0 2022-12-07 07:54:03,942 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8229, 1.8610, 3.9513, 3.6251, 3.7256, 3.7932, 3.1851, 4.0008], device='cuda:1'), covar=tensor([0.1213, 0.1091, 0.0063, 0.0131, 0.0093, 0.0090, 0.0188, 0.0068], device='cuda:1'), in_proj_covar=tensor([0.0119, 0.0132, 0.0070, 0.0095, 0.0080, 0.0085, 0.0066, 0.0068], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:1') 2022-12-07 07:54:17,745 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 07:54:32,248 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7901, 4.7716, 4.3965, 4.9744, 4.6611, 3.7767, 5.0769, 4.9072], device='cuda:1'), covar=tensor([0.0713, 0.0366, 0.0558, 0.0516, 0.0526, 0.0484, 0.0448, 0.0582], device='cuda:1'), in_proj_covar=tensor([0.0095, 0.0073, 0.0096, 0.0092, 0.0100, 0.0065, 0.0091, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 07:54:33,171 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8596, 3.6365, 3.8684, 3.2331, 3.7195, 3.7447, 1.5428, 3.6283], device='cuda:1'), covar=tensor([0.0146, 0.0303, 0.0453, 0.0402, 0.0301, 0.0329, 0.2855, 0.0258], device='cuda:1'), in_proj_covar=tensor([0.0101, 0.0106, 0.0100, 0.0081, 0.0136, 0.0091, 0.0142, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 07:54:43,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.964e+01 3.256e+02 4.500e+02 5.561e+02 1.186e+03, threshold=8.999e+02, percent-clipped=4.0 2022-12-07 07:55:02,567 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 07:55:29,728 INFO [train.py:873] (1/4) Epoch 2, batch 6400, loss[loss=0.2367, simple_loss=0.2274, pruned_loss=0.123, over 14232.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.2256, pruned_loss=0.1173, over 2019659.93 frames. ], batch size: 99, lr: 2.85e-02, grad_scale: 8.0 2022-12-07 07:56:11,802 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 3.193e+02 4.162e+02 5.343e+02 1.659e+03, threshold=8.325e+02, percent-clipped=5.0 2022-12-07 07:56:20,094 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 07:56:57,032 INFO [train.py:873] (1/4) Epoch 2, batch 6500, loss[loss=0.2623, simple_loss=0.2429, pruned_loss=0.1409, over 14323.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.2259, pruned_loss=0.1178, over 2031645.23 frames. ], batch size: 60, lr: 2.84e-02, grad_scale: 8.0 2022-12-07 07:57:04,694 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:57:23,624 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:57:31,990 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:57:38,491 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 3.036e+02 3.911e+02 5.054e+02 1.167e+03, threshold=7.822e+02, percent-clipped=3.0 2022-12-07 07:57:46,666 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:57:54,423 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3840, 4.7795, 4.7763, 5.2387, 4.8839, 4.6021, 5.2936, 4.5903], device='cuda:1'), covar=tensor([0.0228, 0.0864, 0.0216, 0.0345, 0.0626, 0.0349, 0.0403, 0.0396], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0155, 0.0097, 0.0094, 0.0100, 0.0094, 0.0145, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 07:57:58,239 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:58:23,887 INFO [train.py:873] (1/4) Epoch 2, batch 6600, loss[loss=0.2356, simple_loss=0.2322, pruned_loss=0.1195, over 14219.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.2252, pruned_loss=0.1178, over 2004672.30 frames. ], batch size: 60, lr: 2.83e-02, grad_scale: 8.0 2022-12-07 07:58:24,860 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:58:39,675 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:59:06,278 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.900e+02 3.866e+02 4.857e+02 9.306e+02, threshold=7.732e+02, percent-clipped=4.0 2022-12-07 07:59:11,112 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1799, 1.1189, 0.8860, 1.0749, 0.9402, 1.0194, 0.5288, 1.0593], device='cuda:1'), covar=tensor([0.0210, 0.0345, 0.0349, 0.0207, 0.0243, 0.0126, 0.0160, 0.0337], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0013, 0.0013, 0.0013, 0.0015, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([2.6740e-05, 2.7071e-05, 3.1929e-05, 2.8264e-05, 2.9265e-05, 3.1153e-05, 3.4899e-05, 3.2660e-05], device='cuda:1') 2022-12-07 07:59:36,673 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 07:59:51,416 INFO [train.py:873] (1/4) Epoch 2, batch 6700, loss[loss=0.171, simple_loss=0.1556, pruned_loss=0.0932, over 1260.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.2255, pruned_loss=0.1177, over 1990309.69 frames. ], batch size: 100, lr: 2.82e-02, grad_scale: 8.0 2022-12-07 08:00:13,940 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9890, 2.9267, 2.5444, 2.3966, 2.9221, 2.9636, 3.0369, 2.9098], device='cuda:1'), covar=tensor([0.0887, 0.0916, 0.1868, 0.2579, 0.0805, 0.0848, 0.1248, 0.1329], device='cuda:1'), in_proj_covar=tensor([0.0159, 0.0152, 0.0195, 0.0232, 0.0161, 0.0180, 0.0187, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:00:32,374 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 3.050e+02 4.021e+02 5.480e+02 1.335e+03, threshold=8.042e+02, percent-clipped=5.0 2022-12-07 08:01:18,007 INFO [train.py:873] (1/4) Epoch 2, batch 6800, loss[loss=0.2358, simple_loss=0.198, pruned_loss=0.1367, over 1261.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.225, pruned_loss=0.1172, over 1975794.78 frames. ], batch size: 100, lr: 2.81e-02, grad_scale: 8.0 2022-12-07 08:01:43,681 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:01:51,994 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3474, 0.7056, 0.8573, 0.8217, 1.0082, 0.5262, 1.1594, 1.0892], device='cuda:1'), covar=tensor([0.0486, 0.0504, 0.0293, 0.1029, 0.0807, 0.0246, 0.0354, 0.0231], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0018, 0.0020, 0.0017, 0.0017, 0.0022, 0.0015, 0.0015], device='cuda:1'), out_proj_covar=tensor([4.3517e-05, 4.7242e-05, 4.4048e-05, 4.4921e-05, 4.1051e-05, 4.8599e-05, 4.0983e-05, 3.8748e-05], device='cuda:1') 2022-12-07 08:01:59,489 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 3.284e+02 4.234e+02 5.689e+02 1.698e+03, threshold=8.469e+02, percent-clipped=9.0 2022-12-07 08:02:01,543 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:14,269 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:25,182 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:41,243 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:44,542 INFO [train.py:873] (1/4) Epoch 2, batch 6900, loss[loss=0.2673, simple_loss=0.2538, pruned_loss=0.1404, over 14511.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.225, pruned_loss=0.1171, over 1948721.15 frames. ], batch size: 22, lr: 2.80e-02, grad_scale: 8.0 2022-12-07 08:02:48,100 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0009, 1.9322, 4.1326, 2.8186, 3.9055, 1.9347, 3.3296, 3.7563], device='cuda:1'), covar=tensor([0.0191, 0.3614, 0.0126, 0.5473, 0.0103, 0.2867, 0.0526, 0.0131], device='cuda:1'), in_proj_covar=tensor([0.0217, 0.0316, 0.0164, 0.0423, 0.0156, 0.0329, 0.0274, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:02:53,999 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:54,779 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2125, 1.2125, 0.8712, 1.2968, 1.0197, 0.5079, 0.9388, 0.9706], device='cuda:1'), covar=tensor([0.0492, 0.0182, 0.0236, 0.0134, 0.0495, 0.0192, 0.0219, 0.0239], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0018, 0.0020, 0.0016, 0.0016, 0.0021, 0.0015, 0.0015], device='cuda:1'), out_proj_covar=tensor([4.3240e-05, 4.5941e-05, 4.4124e-05, 4.3492e-05, 4.0059e-05, 4.8047e-05, 4.1764e-05, 3.8285e-05], device='cuda:1') 2022-12-07 08:02:55,549 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 08:03:05,485 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:03:08,862 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0198, 1.9256, 1.9374, 2.0192, 1.9698, 1.9890, 1.1473, 1.8082], device='cuda:1'), covar=tensor([0.0304, 0.0402, 0.0536, 0.0198, 0.0362, 0.0395, 0.1793, 0.0429], device='cuda:1'), in_proj_covar=tensor([0.0100, 0.0103, 0.0098, 0.0081, 0.0133, 0.0092, 0.0140, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 08:03:25,508 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 3.007e+02 4.167e+02 5.148e+02 1.275e+03, threshold=8.334e+02, percent-clipped=3.0 2022-12-07 08:03:53,918 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2022-12-07 08:03:57,839 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:04:11,092 INFO [train.py:873] (1/4) Epoch 2, batch 7000, loss[loss=0.2109, simple_loss=0.2184, pruned_loss=0.1017, over 14355.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.2236, pruned_loss=0.1156, over 1988779.26 frames. ], batch size: 73, lr: 2.79e-02, grad_scale: 8.0 2022-12-07 08:04:53,422 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 3.149e+02 3.940e+02 5.259e+02 1.451e+03, threshold=7.881e+02, percent-clipped=6.0 2022-12-07 08:05:37,945 INFO [train.py:873] (1/4) Epoch 2, batch 7100, loss[loss=0.228, simple_loss=0.195, pruned_loss=0.1305, over 2671.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.2228, pruned_loss=0.115, over 1972688.44 frames. ], batch size: 100, lr: 2.79e-02, grad_scale: 16.0 2022-12-07 08:05:38,946 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6528, 4.1774, 4.2264, 4.5917, 4.4879, 4.1497, 4.5688, 4.0381], device='cuda:1'), covar=tensor([0.0289, 0.0869, 0.0276, 0.0470, 0.0513, 0.0528, 0.0603, 0.0474], device='cuda:1'), in_proj_covar=tensor([0.0092, 0.0161, 0.0102, 0.0097, 0.0099, 0.0097, 0.0148, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:06:19,126 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.591e+01 3.151e+02 3.961e+02 5.034e+02 1.065e+03, threshold=7.923e+02, percent-clipped=3.0 2022-12-07 08:06:34,304 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:06:44,685 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:00,688 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:04,456 INFO [train.py:873] (1/4) Epoch 2, batch 7200, loss[loss=0.2181, simple_loss=0.219, pruned_loss=0.1086, over 13871.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.2226, pruned_loss=0.115, over 1997062.85 frames. ], batch size: 23, lr: 2.78e-02, grad_scale: 16.0 2022-12-07 08:07:09,550 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:15,476 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:15,581 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:07:21,392 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0663, 1.7906, 2.0013, 2.0568, 2.1160, 2.0212, 2.1353, 1.8283], device='cuda:1'), covar=tensor([0.0495, 0.1171, 0.0496, 0.0594, 0.0549, 0.0565, 0.0761, 0.0696], device='cuda:1'), in_proj_covar=tensor([0.0094, 0.0161, 0.0104, 0.0098, 0.0101, 0.0097, 0.0150, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:07:37,073 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:37,349 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-07 08:07:38,825 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6101, 2.1233, 2.0199, 1.2628, 2.2246, 2.1829, 2.5352, 2.1044], device='cuda:1'), covar=tensor([0.0658, 0.6080, 0.1222, 0.2930, 0.0829, 0.0530, 0.2451, 0.1323], device='cuda:1'), in_proj_covar=tensor([0.0064, 0.0179, 0.0088, 0.0117, 0.0070, 0.0075, 0.0067, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:07:42,113 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:45,740 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.988e+02 3.907e+02 5.109e+02 1.521e+03, threshold=7.813e+02, percent-clipped=5.0 2022-12-07 08:07:56,606 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:08:12,980 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:08:19,973 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3543, 2.4550, 3.5816, 2.5613, 3.3384, 3.1981, 3.2634, 2.7409], device='cuda:1'), covar=tensor([0.0092, 0.0788, 0.0049, 0.0448, 0.0133, 0.0151, 0.0297, 0.0796], device='cuda:1'), in_proj_covar=tensor([0.0190, 0.0314, 0.0142, 0.0251, 0.0197, 0.0204, 0.0212, 0.0346], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:08:29,763 INFO [train.py:873] (1/4) Epoch 2, batch 7300, loss[loss=0.2382, simple_loss=0.2275, pruned_loss=0.1244, over 14292.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.2233, pruned_loss=0.1155, over 1981624.26 frames. ], batch size: 76, lr: 2.77e-02, grad_scale: 16.0 2022-12-07 08:08:47,720 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1326, 1.3998, 0.8019, 0.8872, 1.1311, 1.1228, 1.1152, 1.2292], device='cuda:1'), covar=tensor([0.1157, 0.1272, 0.1899, 0.2725, 0.1320, 0.0278, 0.0298, 0.0760], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0013, 0.0013, 0.0012, 0.0014, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([2.8114e-05, 2.8308e-05, 3.3822e-05, 2.9077e-05, 2.9458e-05, 3.0741e-05, 3.8143e-05, 3.4715e-05], device='cuda:1') 2022-12-07 08:08:59,280 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2046, 0.7169, 0.9738, 1.1864, 1.3132, 0.3818, 1.2030, 1.2175], device='cuda:1'), covar=tensor([0.0280, 0.0298, 0.0134, 0.0179, 0.0543, 0.0160, 0.0103, 0.0220], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0019, 0.0015, 0.0016, 0.0020, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([4.1703e-05, 4.2649e-05, 4.3369e-05, 4.0874e-05, 3.9518e-05, 4.7169e-05, 3.6946e-05, 3.7311e-05], device='cuda:1') 2022-12-07 08:09:06,225 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 08:09:10,699 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.806e+02 3.909e+02 5.179e+02 1.856e+03, threshold=7.818e+02, percent-clipped=6.0 2022-12-07 08:09:23,614 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8268, 1.0278, 0.9995, 1.0901, 0.7077, 1.1055, 0.8499, 0.7487], device='cuda:1'), covar=tensor([0.3032, 0.0826, 0.0533, 0.0422, 0.1152, 0.0377, 0.1541, 0.1647], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0064, 0.0053, 0.0057, 0.0063, 0.0051, 0.0064, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:09:55,278 INFO [train.py:873] (1/4) Epoch 2, batch 7400, loss[loss=0.2171, simple_loss=0.2195, pruned_loss=0.1074, over 14269.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.2229, pruned_loss=0.115, over 2000008.35 frames. ], batch size: 28, lr: 2.76e-02, grad_scale: 16.0 2022-12-07 08:10:39,741 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.247e+01 3.005e+02 3.840e+02 4.787e+02 1.067e+03, threshold=7.681e+02, percent-clipped=4.0 2022-12-07 08:10:45,410 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7965, 1.5125, 3.7028, 3.5480, 3.6859, 3.7001, 3.0617, 3.8098], device='cuda:1'), covar=tensor([0.1446, 0.1523, 0.0109, 0.0131, 0.0126, 0.0122, 0.0170, 0.0111], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0135, 0.0072, 0.0098, 0.0083, 0.0088, 0.0066, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:1') 2022-12-07 08:11:22,118 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4181, 3.0943, 4.5704, 4.2042, 4.3856, 3.0585, 4.2310, 3.0376], device='cuda:1'), covar=tensor([0.0037, 0.0123, 0.0066, 0.0077, 0.0043, 0.0239, 0.0019, 0.0193], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0138, 0.0152, 0.0138, 0.0120, 0.0181, 0.0085, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:1') 2022-12-07 08:11:24,371 INFO [train.py:873] (1/4) Epoch 2, batch 7500, loss[loss=0.2594, simple_loss=0.2321, pruned_loss=0.1434, over 7748.00 frames. ], tot_loss[loss=0.227, simple_loss=0.2231, pruned_loss=0.1154, over 2018472.43 frames. ], batch size: 100, lr: 2.75e-02, grad_scale: 8.0 2022-12-07 08:11:29,650 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:11:38,383 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-07 08:11:52,683 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:12:03,099 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.166e+01 2.953e+02 4.080e+02 5.440e+02 1.285e+03, threshold=8.159e+02, percent-clipped=6.0 2022-12-07 08:12:04,017 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 08:12:04,289 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.94 vs. limit=2.0 2022-12-07 08:12:04,892 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.66 vs. limit=2.0 2022-12-07 08:12:05,497 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:12:07,038 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=10.18 vs. limit=5.0 2022-12-07 08:12:48,570 INFO [train.py:873] (1/4) Epoch 3, batch 0, loss[loss=0.2861, simple_loss=0.2595, pruned_loss=0.1563, over 6939.00 frames. ], tot_loss[loss=0.2861, simple_loss=0.2595, pruned_loss=0.1563, over 6939.00 frames. ], batch size: 100, lr: 2.61e-02, grad_scale: 8.0 2022-12-07 08:12:48,570 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 08:12:55,593 INFO [train.py:905] (1/4) Epoch 3, validation: loss=0.159, simple_loss=0.1997, pruned_loss=0.05911, over 857387.00 frames. 2022-12-07 08:12:55,594 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 08:13:12,551 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:13:14,638 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 08:13:53,451 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:13:53,897 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 08:13:57,127 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:14:11,334 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 4.526e+01 3.048e+02 3.683e+02 5.299e+02 1.167e+03, threshold=7.365e+02, percent-clipped=4.0 2022-12-07 08:14:23,259 INFO [train.py:873] (1/4) Epoch 3, batch 100, loss[loss=0.2015, simple_loss=0.2097, pruned_loss=0.09663, over 14095.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.2241, pruned_loss=0.1154, over 860089.13 frames. ], batch size: 22, lr: 2.60e-02, grad_scale: 8.0 2022-12-07 08:14:31,192 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2022-12-07 08:14:49,266 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 08:15:33,436 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 08:15:38,763 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 3.082e+02 4.015e+02 5.606e+02 9.270e+02, threshold=8.031e+02, percent-clipped=6.0 2022-12-07 08:15:49,797 INFO [train.py:873] (1/4) Epoch 3, batch 200, loss[loss=0.21, simple_loss=0.1754, pruned_loss=0.1223, over 1258.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2212, pruned_loss=0.1124, over 1325193.91 frames. ], batch size: 100, lr: 2.59e-02, grad_scale: 8.0 2022-12-07 08:16:10,909 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 08:16:13,170 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 08:16:28,012 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-07 08:16:48,541 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0265, 3.7081, 4.8853, 3.8829, 4.8313, 4.6902, 4.4316, 4.2415], device='cuda:1'), covar=tensor([0.0048, 0.0777, 0.0037, 0.0360, 0.0126, 0.0079, 0.0493, 0.0617], device='cuda:1'), in_proj_covar=tensor([0.0198, 0.0317, 0.0149, 0.0254, 0.0202, 0.0207, 0.0221, 0.0354], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:16:50,919 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:17:04,341 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 08:17:04,617 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.006e+02 3.006e+02 3.776e+02 5.007e+02 1.251e+03, threshold=7.551e+02, percent-clipped=5.0 2022-12-07 08:17:16,243 INFO [train.py:873] (1/4) Epoch 3, batch 300, loss[loss=0.218, simple_loss=0.2202, pruned_loss=0.1079, over 14031.00 frames. ], tot_loss[loss=0.222, simple_loss=0.2204, pruned_loss=0.1118, over 1655041.70 frames. ], batch size: 22, lr: 2.59e-02, grad_scale: 8.0 2022-12-07 08:17:19,226 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.15 vs. limit=2.0 2022-12-07 08:17:32,196 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:18:16,406 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 08:18:31,036 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.693e+02 3.512e+02 4.941e+02 9.484e+02, threshold=7.024e+02, percent-clipped=2.0 2022-12-07 08:18:41,946 INFO [train.py:873] (1/4) Epoch 3, batch 400, loss[loss=0.215, simple_loss=0.2309, pruned_loss=0.0996, over 14250.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.22, pruned_loss=0.1102, over 1790431.05 frames. ], batch size: 37, lr: 2.58e-02, grad_scale: 8.0 2022-12-07 08:19:03,972 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:19:48,985 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 08:19:56,022 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 3.036e+02 3.993e+02 5.394e+02 1.334e+03, threshold=7.985e+02, percent-clipped=7.0 2022-12-07 08:20:07,948 INFO [train.py:873] (1/4) Epoch 3, batch 500, loss[loss=0.2491, simple_loss=0.2471, pruned_loss=0.1255, over 14088.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.2202, pruned_loss=0.112, over 1805270.21 frames. ], batch size: 29, lr: 2.57e-02, grad_scale: 8.0 2022-12-07 08:20:08,370 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 08:21:06,248 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 08:21:17,398 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1558, 2.0415, 3.8725, 2.8433, 3.9939, 2.1202, 3.1237, 3.9700], device='cuda:1'), covar=tensor([0.0170, 0.3982, 0.0209, 0.5772, 0.0138, 0.2818, 0.0784, 0.0087], device='cuda:1'), in_proj_covar=tensor([0.0214, 0.0325, 0.0176, 0.0432, 0.0156, 0.0338, 0.0287, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:21:22,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.935e+02 3.905e+02 4.916e+02 1.140e+03, threshold=7.811e+02, percent-clipped=3.0 2022-12-07 08:21:32,702 INFO [train.py:873] (1/4) Epoch 3, batch 600, loss[loss=0.2132, simple_loss=0.1957, pruned_loss=0.1153, over 3922.00 frames. ], tot_loss[loss=0.2206, simple_loss=0.219, pruned_loss=0.1111, over 1891770.10 frames. ], batch size: 100, lr: 2.56e-02, grad_scale: 8.0 2022-12-07 08:22:01,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5189, 2.6071, 3.6245, 2.8632, 3.3758, 3.3089, 3.4015, 2.9750], device='cuda:1'), covar=tensor([0.0074, 0.0790, 0.0056, 0.0381, 0.0128, 0.0126, 0.0332, 0.0662], device='cuda:1'), in_proj_covar=tensor([0.0202, 0.0328, 0.0154, 0.0263, 0.0205, 0.0217, 0.0227, 0.0360], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:22:47,078 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 3.132e+02 3.850e+02 5.057e+02 1.369e+03, threshold=7.700e+02, percent-clipped=5.0 2022-12-07 08:22:58,912 INFO [train.py:873] (1/4) Epoch 3, batch 700, loss[loss=0.2096, simple_loss=0.2176, pruned_loss=0.1008, over 14561.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2189, pruned_loss=0.1109, over 1961833.59 frames. ], batch size: 34, lr: 2.56e-02, grad_scale: 8.0 2022-12-07 08:23:20,519 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:23:48,053 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:23:55,424 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2008, 3.9153, 3.8099, 4.2321, 4.1809, 3.7187, 4.2552, 3.6563], device='cuda:1'), covar=tensor([0.0377, 0.0832, 0.0283, 0.0378, 0.0510, 0.0807, 0.0432, 0.0499], device='cuda:1'), in_proj_covar=tensor([0.0092, 0.0164, 0.0102, 0.0100, 0.0098, 0.0093, 0.0149, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:24:01,244 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:24:12,922 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 3.132e+02 4.108e+02 5.472e+02 1.235e+03, threshold=8.216e+02, percent-clipped=5.0 2022-12-07 08:24:23,975 INFO [train.py:873] (1/4) Epoch 3, batch 800, loss[loss=0.2089, simple_loss=0.2135, pruned_loss=0.1021, over 14263.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.2177, pruned_loss=0.1098, over 1981844.30 frames. ], batch size: 63, lr: 2.55e-02, grad_scale: 8.0 2022-12-07 08:24:40,915 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:24:51,440 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1163, 2.9011, 3.1882, 2.9844, 2.9873, 2.5513, 1.2133, 2.9626], device='cuda:1'), covar=tensor([0.0229, 0.0426, 0.0431, 0.0325, 0.0378, 0.0918, 0.3043, 0.0250], device='cuda:1'), in_proj_covar=tensor([0.0105, 0.0107, 0.0101, 0.0084, 0.0140, 0.0096, 0.0144, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 08:25:30,394 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:25:39,382 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.828e+02 4.017e+02 4.886e+02 9.739e+02, threshold=8.033e+02, percent-clipped=2.0 2022-12-07 08:25:43,883 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8229, 1.5879, 2.1360, 1.6778, 2.1445, 1.6767, 1.7517, 1.8148], device='cuda:1'), covar=tensor([0.0059, 0.0220, 0.0025, 0.0054, 0.0036, 0.0077, 0.0044, 0.0107], device='cuda:1'), in_proj_covar=tensor([0.0198, 0.0327, 0.0156, 0.0258, 0.0208, 0.0216, 0.0227, 0.0352], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:25:45,559 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5180, 1.3652, 3.3317, 1.4971, 3.3570, 3.3477, 2.2526, 3.6814], device='cuda:1'), covar=tensor([0.0165, 0.2692, 0.0301, 0.2210, 0.0399, 0.0294, 0.0924, 0.0153], device='cuda:1'), in_proj_covar=tensor([0.0111, 0.0145, 0.0100, 0.0153, 0.0120, 0.0106, 0.0096, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:25:51,107 INFO [train.py:873] (1/4) Epoch 3, batch 900, loss[loss=0.1773, simple_loss=0.1553, pruned_loss=0.09961, over 1249.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.2186, pruned_loss=0.1102, over 1999509.54 frames. ], batch size: 100, lr: 2.54e-02, grad_scale: 8.0 2022-12-07 08:26:22,570 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:26:41,979 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2951, 3.8498, 3.7835, 4.2688, 4.1168, 3.9218, 4.3117, 3.6879], device='cuda:1'), covar=tensor([0.0338, 0.0894, 0.0312, 0.0415, 0.0619, 0.0636, 0.0498, 0.0518], device='cuda:1'), in_proj_covar=tensor([0.0092, 0.0163, 0.0101, 0.0100, 0.0098, 0.0094, 0.0151, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:26:42,924 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:26:46,157 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7473, 3.2213, 4.5447, 3.2667, 4.3280, 4.4105, 4.0024, 3.8131], device='cuda:1'), covar=tensor([0.0080, 0.0954, 0.0058, 0.0525, 0.0167, 0.0141, 0.0864, 0.0671], device='cuda:1'), in_proj_covar=tensor([0.0198, 0.0328, 0.0155, 0.0259, 0.0207, 0.0214, 0.0227, 0.0348], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:27:06,256 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 2.979e+02 3.878e+02 4.942e+02 1.281e+03, threshold=7.755e+02, percent-clipped=6.0 2022-12-07 08:27:16,836 INFO [train.py:873] (1/4) Epoch 3, batch 1000, loss[loss=0.2308, simple_loss=0.2246, pruned_loss=0.1185, over 14236.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.2196, pruned_loss=0.1116, over 1955817.08 frames. ], batch size: 60, lr: 2.54e-02, grad_scale: 8.0 2022-12-07 08:27:35,226 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:27:38,024 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:28:09,741 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0453, 0.5259, 1.0514, 1.0970, 1.0970, 0.5221, 1.4112, 0.9192], device='cuda:1'), covar=tensor([0.0472, 0.0491, 0.0171, 0.0414, 0.0721, 0.0192, 0.0714, 0.0480], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0017, 0.0018, 0.0015, 0.0018, 0.0021, 0.0015, 0.0015], device='cuda:1'), out_proj_covar=tensor([4.4994e-05, 4.4925e-05, 4.5651e-05, 4.3830e-05, 4.4304e-05, 5.2755e-05, 4.4813e-05, 3.8655e-05], device='cuda:1') 2022-12-07 08:28:19,063 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 08:28:29,939 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:28:31,352 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 3.033e+02 4.056e+02 5.147e+02 1.062e+03, threshold=8.113e+02, percent-clipped=5.0 2022-12-07 08:28:42,881 INFO [train.py:873] (1/4) Epoch 3, batch 1100, loss[loss=0.215, simple_loss=0.1862, pruned_loss=0.1219, over 2602.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.2193, pruned_loss=0.1123, over 1918409.07 frames. ], batch size: 100, lr: 2.53e-02, grad_scale: 8.0 2022-12-07 08:28:55,077 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:29:20,364 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2144, 5.0953, 4.7829, 5.2272, 5.0164, 4.1812, 5.4558, 5.3076], device='cuda:1'), covar=tensor([0.0654, 0.0436, 0.0684, 0.0749, 0.0603, 0.0467, 0.0483, 0.0670], device='cuda:1'), in_proj_covar=tensor([0.0098, 0.0077, 0.0100, 0.0095, 0.0106, 0.0068, 0.0098, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:29:37,239 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 08:29:41,048 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5512, 2.3183, 2.2310, 1.3425, 2.0820, 2.1236, 2.2634, 1.8283], device='cuda:1'), covar=tensor([0.0409, 0.3349, 0.0893, 0.2493, 0.0764, 0.0445, 0.1682, 0.1480], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0196, 0.0095, 0.0121, 0.0079, 0.0078, 0.0070, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:29:41,388 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 08:29:45,977 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2638, 3.0842, 2.8061, 2.1846, 2.5282, 2.9406, 3.0316, 2.5354], device='cuda:1'), covar=tensor([0.0302, 0.2680, 0.0709, 0.1508, 0.0570, 0.0260, 0.1484, 0.0913], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0196, 0.0095, 0.0121, 0.0079, 0.0079, 0.0070, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:29:58,310 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.546e+02 3.661e+02 5.045e+02 9.524e+02, threshold=7.322e+02, percent-clipped=3.0 2022-12-07 08:30:09,551 INFO [train.py:873] (1/4) Epoch 3, batch 1200, loss[loss=0.2417, simple_loss=0.2309, pruned_loss=0.1262, over 14193.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.2187, pruned_loss=0.1109, over 1983860.83 frames. ], batch size: 89, lr: 2.52e-02, grad_scale: 8.0 2022-12-07 08:30:22,511 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 08:30:37,130 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:30:49,001 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:30:59,708 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 08:31:23,706 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 3.060e+02 3.691e+02 4.545e+02 9.377e+02, threshold=7.381e+02, percent-clipped=3.0 2022-12-07 08:31:35,382 INFO [train.py:873] (1/4) Epoch 3, batch 1300, loss[loss=0.2432, simple_loss=0.2382, pruned_loss=0.124, over 14543.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2187, pruned_loss=0.1109, over 1987097.19 frames. ], batch size: 43, lr: 2.51e-02, grad_scale: 8.0 2022-12-07 08:31:41,474 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:31:49,050 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:31:49,629 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 08:31:51,148 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2022-12-07 08:31:52,418 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9396, 2.7684, 2.4993, 2.4874, 2.7596, 2.7227, 2.8244, 2.8886], device='cuda:1'), covar=tensor([0.0677, 0.0608, 0.1391, 0.1988, 0.0699, 0.0663, 0.0928, 0.0842], device='cuda:1'), in_proj_covar=tensor([0.0182, 0.0166, 0.0224, 0.0274, 0.0183, 0.0200, 0.0222, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:31:55,975 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7273, 3.5234, 3.9549, 3.4198, 3.6998, 3.6764, 1.3685, 3.6667], device='cuda:1'), covar=tensor([0.0188, 0.0327, 0.0341, 0.0460, 0.0317, 0.0315, 0.3090, 0.0217], device='cuda:1'), in_proj_covar=tensor([0.0104, 0.0109, 0.0104, 0.0087, 0.0143, 0.0095, 0.0143, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 08:31:57,618 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4071, 2.0148, 1.8429, 2.0459, 1.3048, 1.6434, 1.9858, 1.1105], device='cuda:1'), covar=tensor([0.3501, 0.1182, 0.2413, 0.1286, 0.1195, 0.1049, 0.1215, 0.2265], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0061, 0.0049, 0.0058, 0.0065, 0.0050, 0.0068, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:32:11,271 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 08:32:29,374 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7725, 3.5069, 3.3618, 3.3211, 3.5804, 3.6123, 3.7653, 3.7039], device='cuda:1'), covar=tensor([0.0668, 0.0733, 0.1160, 0.2026, 0.0610, 0.0517, 0.0751, 0.0926], device='cuda:1'), in_proj_covar=tensor([0.0187, 0.0168, 0.0223, 0.0277, 0.0185, 0.0197, 0.0220, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:32:32,634 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9374, 1.2451, 0.7864, 0.9817, 0.7928, 0.9430, 0.6117, 0.7408], device='cuda:1'), covar=tensor([0.0286, 0.0215, 0.0339, 0.0523, 0.0509, 0.0211, 0.0239, 0.0441], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0011, 0.0012, 0.0012, 0.0013, 0.0015, 0.0014, 0.0015], device='cuda:1'), out_proj_covar=tensor([3.1954e-05, 2.7847e-05, 3.2491e-05, 3.0123e-05, 3.2002e-05, 3.3791e-05, 4.0207e-05, 3.9692e-05], device='cuda:1') 2022-12-07 08:32:35,158 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7985, 2.1859, 2.0451, 2.1357, 1.4275, 1.8704, 2.2778, 1.4171], device='cuda:1'), covar=tensor([0.2737, 0.1271, 0.2366, 0.2611, 0.1136, 0.0677, 0.1624, 0.1812], device='cuda:1'), in_proj_covar=tensor([0.0158, 0.0060, 0.0050, 0.0058, 0.0065, 0.0049, 0.0068, 0.0081], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:32:43,980 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:32:49,717 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.808e+02 3.737e+02 4.916e+02 1.118e+03, threshold=7.474e+02, percent-clipped=8.0 2022-12-07 08:33:01,054 INFO [train.py:873] (1/4) Epoch 3, batch 1400, loss[loss=0.171, simple_loss=0.1535, pruned_loss=0.09422, over 1237.00 frames. ], tot_loss[loss=0.221, simple_loss=0.2191, pruned_loss=0.1114, over 1997531.37 frames. ], batch size: 100, lr: 2.51e-02, grad_scale: 8.0 2022-12-07 08:33:12,616 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:33:54,101 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:34:15,420 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.141e+01 3.095e+02 4.057e+02 5.380e+02 9.685e+02, threshold=8.113e+02, percent-clipped=8.0 2022-12-07 08:34:15,558 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1046, 1.2200, 1.4841, 0.7951, 1.1442, 1.3241, 0.8309, 1.2425], device='cuda:1'), covar=tensor([0.1008, 0.0920, 0.0200, 0.1331, 0.1143, 0.0355, 0.1573, 0.0644], device='cuda:1'), in_proj_covar=tensor([0.0060, 0.0063, 0.0056, 0.0066, 0.0074, 0.0060, 0.0116, 0.0068], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 08:34:27,198 INFO [train.py:873] (1/4) Epoch 3, batch 1500, loss[loss=0.2402, simple_loss=0.2239, pruned_loss=0.1283, over 8602.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2186, pruned_loss=0.11, over 2035668.73 frames. ], batch size: 100, lr: 2.50e-02, grad_scale: 8.0 2022-12-07 08:34:35,200 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5064, 3.2655, 2.3171, 3.6500, 2.9987, 3.4090, 3.1690, 2.2849], device='cuda:1'), covar=tensor([0.0157, 0.0376, 0.1901, 0.0120, 0.0235, 0.0254, 0.0412, 0.2331], device='cuda:1'), in_proj_covar=tensor([0.0194, 0.0259, 0.0343, 0.0157, 0.0188, 0.0186, 0.0240, 0.0349], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:34:37,672 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5139, 3.0708, 3.9698, 2.6281, 2.7227, 2.6140, 0.9685, 2.7294], device='cuda:1'), covar=tensor([0.1728, 0.0632, 0.0489, 0.1039, 0.0982, 0.1474, 0.3832, 0.1311], device='cuda:1'), in_proj_covar=tensor([0.0062, 0.0064, 0.0058, 0.0068, 0.0076, 0.0061, 0.0118, 0.0069], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 08:34:54,224 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:35:35,806 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:35:41,765 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.784e+02 3.654e+02 4.523e+02 7.286e+02, threshold=7.308e+02, percent-clipped=0.0 2022-12-07 08:35:52,814 INFO [train.py:873] (1/4) Epoch 3, batch 1600, loss[loss=0.2679, simple_loss=0.2361, pruned_loss=0.1499, over 7804.00 frames. ], tot_loss[loss=0.218, simple_loss=0.2177, pruned_loss=0.1092, over 2024044.57 frames. ], batch size: 100, lr: 2.49e-02, grad_scale: 8.0 2022-12-07 08:35:54,529 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:36:06,511 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:36:12,942 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0579, 2.3984, 2.2634, 2.2024, 1.7488, 2.0138, 2.6281, 1.6779], device='cuda:1'), covar=tensor([0.3228, 0.1663, 0.3301, 0.4765, 0.1202, 0.1515, 0.1709, 0.2177], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0064, 0.0051, 0.0059, 0.0064, 0.0050, 0.0071, 0.0082], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:36:48,046 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:01,894 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:07,697 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 3.316e+02 4.284e+02 5.393e+02 9.437e+02, threshold=8.568e+02, percent-clipped=8.0 2022-12-07 08:37:19,029 INFO [train.py:873] (1/4) Epoch 3, batch 1700, loss[loss=0.2008, simple_loss=0.2054, pruned_loss=0.09815, over 14450.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2183, pruned_loss=0.1089, over 2130354.87 frames. ], batch size: 53, lr: 2.49e-02, grad_scale: 8.0 2022-12-07 08:37:26,011 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9738, 1.7091, 2.2511, 1.3744, 1.7069, 1.9773, 1.1298, 1.9820], device='cuda:1'), covar=tensor([0.0751, 0.1051, 0.0429, 0.1732, 0.1182, 0.0720, 0.2658, 0.0794], device='cuda:1'), in_proj_covar=tensor([0.0062, 0.0066, 0.0062, 0.0071, 0.0075, 0.0063, 0.0119, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 08:37:38,887 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 08:37:42,667 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:55,908 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:38:28,220 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:38:33,243 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.887e+02 3.816e+02 4.798e+02 1.128e+03, threshold=7.633e+02, percent-clipped=3.0 2022-12-07 08:38:44,615 INFO [train.py:873] (1/4) Epoch 3, batch 1800, loss[loss=0.2041, simple_loss=0.2098, pruned_loss=0.0992, over 14293.00 frames. ], tot_loss[loss=0.218, simple_loss=0.2181, pruned_loss=0.109, over 2087200.78 frames. ], batch size: 39, lr: 2.48e-02, grad_scale: 8.0 2022-12-07 08:38:48,256 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:39:21,056 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:39:59,562 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 3.016e+02 3.995e+02 5.163e+02 9.262e+02, threshold=7.990e+02, percent-clipped=4.0 2022-12-07 08:40:11,094 INFO [train.py:873] (1/4) Epoch 3, batch 1900, loss[loss=0.2031, simple_loss=0.1776, pruned_loss=0.1143, over 2677.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2178, pruned_loss=0.1095, over 2032055.25 frames. ], batch size: 100, lr: 2.47e-02, grad_scale: 8.0 2022-12-07 08:40:12,984 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:40:53,617 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:41:00,005 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2022-12-07 08:41:25,312 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.833e+02 3.865e+02 5.052e+02 1.376e+03, threshold=7.730e+02, percent-clipped=3.0 2022-12-07 08:41:36,266 INFO [train.py:873] (1/4) Epoch 3, batch 2000, loss[loss=0.2362, simple_loss=0.2277, pruned_loss=0.1224, over 10318.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.2186, pruned_loss=0.1108, over 1978638.55 frames. ], batch size: 100, lr: 2.47e-02, grad_scale: 16.0 2022-12-07 08:41:54,173 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1128, 2.8733, 2.6272, 2.7434, 2.9474, 2.9361, 3.0520, 3.0642], device='cuda:1'), covar=tensor([0.0535, 0.0636, 0.1319, 0.1672, 0.0618, 0.0530, 0.0863, 0.0654], device='cuda:1'), in_proj_covar=tensor([0.0192, 0.0170, 0.0226, 0.0281, 0.0186, 0.0207, 0.0224, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 08:42:01,071 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2567, 2.9512, 2.4030, 2.0611, 1.5894, 2.1036, 2.8022, 1.5852], device='cuda:1'), covar=tensor([0.3261, 0.1232, 0.1147, 0.4572, 0.1171, 0.1124, 0.1252, 0.2659], device='cuda:1'), in_proj_covar=tensor([0.0160, 0.0068, 0.0051, 0.0061, 0.0066, 0.0054, 0.0075, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:42:04,433 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6209, 2.7103, 1.9278, 2.9260, 2.7065, 2.9070, 2.5508, 2.2720], device='cuda:1'), covar=tensor([0.0130, 0.0350, 0.1562, 0.0106, 0.0151, 0.0131, 0.0435, 0.1488], device='cuda:1'), in_proj_covar=tensor([0.0199, 0.0266, 0.0344, 0.0160, 0.0192, 0.0190, 0.0245, 0.0347], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:42:30,613 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:42:48,459 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2157, 0.8118, 0.7176, 0.9168, 0.8194, 0.5740, 1.1830, 0.9147], device='cuda:1'), covar=tensor([0.0217, 0.0295, 0.0173, 0.0415, 0.0246, 0.0138, 0.0329, 0.0201], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0017, 0.0018, 0.0015, 0.0016, 0.0019, 0.0015, 0.0014], device='cuda:1'), out_proj_covar=tensor([4.2596e-05, 4.6733e-05, 4.6435e-05, 4.5358e-05, 4.2320e-05, 4.9362e-05, 4.3070e-05, 3.7465e-05], device='cuda:1') 2022-12-07 08:42:49,868 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 3.144e+02 4.016e+02 5.229e+02 1.105e+03, threshold=8.031e+02, percent-clipped=5.0 2022-12-07 08:43:00,503 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:43:01,531 INFO [train.py:873] (1/4) Epoch 3, batch 2100, loss[loss=0.203, simple_loss=0.2056, pruned_loss=0.1002, over 14109.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.2174, pruned_loss=0.1095, over 1949326.86 frames. ], batch size: 29, lr: 2.46e-02, grad_scale: 16.0 2022-12-07 08:43:18,100 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2022-12-07 08:43:22,595 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:43:32,875 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:44:01,036 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 08:44:16,384 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.883e+02 3.756e+02 4.821e+02 7.319e+02, threshold=7.512e+02, percent-clipped=0.0 2022-12-07 08:44:27,588 INFO [train.py:873] (1/4) Epoch 3, batch 2200, loss[loss=0.2304, simple_loss=0.2188, pruned_loss=0.121, over 11133.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.218, pruned_loss=0.1101, over 1997941.02 frames. ], batch size: 100, lr: 2.45e-02, grad_scale: 16.0 2022-12-07 08:44:43,604 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2565, 3.8101, 3.8308, 4.2227, 4.1054, 3.8455, 4.2612, 3.7561], device='cuda:1'), covar=tensor([0.0327, 0.0897, 0.0297, 0.0395, 0.0502, 0.0648, 0.0461, 0.0388], device='cuda:1'), in_proj_covar=tensor([0.0101, 0.0169, 0.0109, 0.0103, 0.0101, 0.0097, 0.0156, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:44:59,076 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:45:17,016 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:45:42,085 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 3.126e+02 3.760e+02 4.600e+02 9.821e+02, threshold=7.519e+02, percent-clipped=3.0 2022-12-07 08:45:51,624 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:45:53,185 INFO [train.py:873] (1/4) Epoch 3, batch 2300, loss[loss=0.235, simple_loss=0.2124, pruned_loss=0.1289, over 5011.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.2169, pruned_loss=0.1095, over 1942456.98 frames. ], batch size: 100, lr: 2.45e-02, grad_scale: 16.0 2022-12-07 08:46:09,821 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 08:46:15,686 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5498, 4.9420, 4.9553, 5.5373, 5.3258, 4.6334, 5.4321, 4.5577], device='cuda:1'), covar=tensor([0.0263, 0.0730, 0.0199, 0.0280, 0.0451, 0.0299, 0.0392, 0.0384], device='cuda:1'), in_proj_covar=tensor([0.0102, 0.0171, 0.0110, 0.0104, 0.0104, 0.0097, 0.0159, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 08:47:07,374 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.910e+02 3.962e+02 5.251e+02 8.178e+02, threshold=7.923e+02, percent-clipped=2.0 2022-12-07 08:47:09,119 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2022-12-07 08:47:17,985 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:47:18,683 INFO [train.py:873] (1/4) Epoch 3, batch 2400, loss[loss=0.2244, simple_loss=0.2269, pruned_loss=0.111, over 14217.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2177, pruned_loss=0.1103, over 1951979.86 frames. ], batch size: 94, lr: 2.44e-02, grad_scale: 16.0 2022-12-07 08:47:35,594 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:47:50,035 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:47:58,497 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:47:59,291 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3411, 1.4546, 2.4723, 1.3067, 2.4353, 2.5415, 1.9144, 2.4768], device='cuda:1'), covar=tensor([0.0151, 0.1288, 0.0205, 0.1284, 0.0195, 0.0223, 0.0537, 0.0142], device='cuda:1'), in_proj_covar=tensor([0.0112, 0.0143, 0.0104, 0.0153, 0.0118, 0.0106, 0.0094, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:48:09,291 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2022-12-07 08:48:21,387 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5950, 3.0772, 4.7053, 4.4231, 4.4205, 2.9673, 4.7534, 3.8344], device='cuda:1'), covar=tensor([0.0034, 0.0109, 0.0141, 0.0063, 0.0035, 0.0260, 0.0014, 0.0142], device='cuda:1'), in_proj_covar=tensor([0.0123, 0.0151, 0.0180, 0.0160, 0.0135, 0.0198, 0.0094, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:1') 2022-12-07 08:48:31,674 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:48:34,155 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.796e+02 3.862e+02 5.213e+02 1.653e+03, threshold=7.725e+02, percent-clipped=7.0 2022-12-07 08:48:44,388 INFO [train.py:873] (1/4) Epoch 3, batch 2500, loss[loss=0.2072, simple_loss=0.1993, pruned_loss=0.1075, over 4995.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.2175, pruned_loss=0.11, over 1950637.09 frames. ], batch size: 100, lr: 2.43e-02, grad_scale: 8.0 2022-12-07 08:49:49,059 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:49:59,743 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.968e+02 3.839e+02 5.013e+02 8.291e+02, threshold=7.678e+02, percent-clipped=4.0 2022-12-07 08:50:04,225 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:50:05,312 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9810, 2.4158, 4.2096, 3.9866, 4.0925, 2.6632, 3.9891, 3.2358], device='cuda:1'), covar=tensor([0.0045, 0.0170, 0.0118, 0.0074, 0.0033, 0.0264, 0.0022, 0.0189], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0152, 0.0179, 0.0161, 0.0134, 0.0197, 0.0095, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:1') 2022-12-07 08:50:10,238 INFO [train.py:873] (1/4) Epoch 3, batch 2600, loss[loss=0.2065, simple_loss=0.1802, pruned_loss=0.1164, over 1234.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.2154, pruned_loss=0.1083, over 1920398.91 frames. ], batch size: 100, lr: 2.43e-02, grad_scale: 8.0 2022-12-07 08:50:14,789 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:50:22,017 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 08:50:22,590 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 08:50:33,338 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9216, 1.2738, 2.0058, 1.2706, 2.0865, 2.0900, 1.6018, 2.0478], device='cuda:1'), covar=tensor([0.0150, 0.1060, 0.0203, 0.1063, 0.0196, 0.0216, 0.0557, 0.0200], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0148, 0.0105, 0.0157, 0.0122, 0.0111, 0.0097, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:50:41,075 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:51:07,045 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:51:25,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.813e+02 3.769e+02 4.969e+02 1.379e+03, threshold=7.537e+02, percent-clipped=10.0 2022-12-07 08:51:36,296 INFO [train.py:873] (1/4) Epoch 3, batch 2700, loss[loss=0.2012, simple_loss=0.2077, pruned_loss=0.0974, over 14653.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2148, pruned_loss=0.1075, over 1962495.02 frames. ], batch size: 33, lr: 2.42e-02, grad_scale: 8.0 2022-12-07 08:51:53,241 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:51:54,939 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4665, 1.4582, 3.2917, 1.4460, 3.2415, 3.3703, 2.5786, 3.6274], device='cuda:1'), covar=tensor([0.0146, 0.2374, 0.0281, 0.2049, 0.0423, 0.0337, 0.0618, 0.0144], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0150, 0.0105, 0.0159, 0.0123, 0.0111, 0.0098, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:52:06,897 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 08:52:26,334 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.13 vs. limit=2.0 2022-12-07 08:52:34,445 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:52:51,605 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 3.286e+02 4.021e+02 4.924e+02 9.078e+02, threshold=8.042e+02, percent-clipped=2.0 2022-12-07 08:53:01,818 INFO [train.py:873] (1/4) Epoch 3, batch 2800, loss[loss=0.2214, simple_loss=0.2274, pruned_loss=0.1077, over 14065.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2156, pruned_loss=0.1073, over 2044540.00 frames. ], batch size: 29, lr: 2.41e-02, grad_scale: 8.0 2022-12-07 08:53:05,435 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0507, 2.0030, 1.7771, 1.6546, 1.3980, 1.6865, 1.8436, 0.8806], device='cuda:1'), covar=tensor([0.3929, 0.0996, 0.1809, 0.1340, 0.1348, 0.1093, 0.1740, 0.3364], device='cuda:1'), in_proj_covar=tensor([0.0161, 0.0064, 0.0052, 0.0059, 0.0067, 0.0055, 0.0078, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:53:11,011 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.25 vs. limit=2.0 2022-12-07 08:53:31,742 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:16,684 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.972e+02 3.622e+02 4.500e+02 9.762e+02, threshold=7.243e+02, percent-clipped=2.0 2022-12-07 08:54:21,480 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:24,004 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:27,361 INFO [train.py:873] (1/4) Epoch 3, batch 2900, loss[loss=0.214, simple_loss=0.218, pruned_loss=0.105, over 14025.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2145, pruned_loss=0.1066, over 1963152.35 frames. ], batch size: 29, lr: 2.41e-02, grad_scale: 8.0 2022-12-07 08:54:39,710 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:54:53,300 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:01,982 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:11,106 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 08:55:19,256 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:20,072 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:55:42,573 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 3.066e+02 4.043e+02 5.337e+02 1.703e+03, threshold=8.087e+02, percent-clipped=7.0 2022-12-07 08:55:52,803 INFO [train.py:873] (1/4) Epoch 3, batch 3000, loss[loss=0.2094, simple_loss=0.207, pruned_loss=0.1059, over 14128.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2147, pruned_loss=0.1069, over 2010031.94 frames. ], batch size: 19, lr: 2.40e-02, grad_scale: 8.0 2022-12-07 08:55:52,804 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 08:56:03,704 INFO [train.py:905] (1/4) Epoch 3, validation: loss=0.1337, simple_loss=0.176, pruned_loss=0.04573, over 857387.00 frames. 2022-12-07 08:56:03,704 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 08:56:21,886 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8765, 1.3750, 2.9304, 2.8380, 2.9978, 2.8537, 2.3693, 3.0789], device='cuda:1'), covar=tensor([0.0974, 0.1277, 0.0121, 0.0167, 0.0123, 0.0109, 0.0225, 0.0099], device='cuda:1'), in_proj_covar=tensor([0.0129, 0.0142, 0.0078, 0.0109, 0.0089, 0.0098, 0.0072, 0.0078], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 08:56:45,782 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:56:46,938 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:57:19,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 3.113e+02 4.054e+02 5.125e+02 9.563e+02, threshold=8.108e+02, percent-clipped=4.0 2022-12-07 08:57:30,110 INFO [train.py:873] (1/4) Epoch 3, batch 3100, loss[loss=0.2047, simple_loss=0.1951, pruned_loss=0.1071, over 5988.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2143, pruned_loss=0.1065, over 1991023.48 frames. ], batch size: 100, lr: 2.40e-02, grad_scale: 8.0 2022-12-07 08:57:38,183 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 08:57:38,723 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:57:39,604 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:58:03,824 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4893, 3.0528, 2.2764, 3.6195, 3.3621, 3.6441, 2.9647, 2.2311], device='cuda:1'), covar=tensor([0.0198, 0.0556, 0.2306, 0.0148, 0.0226, 0.0271, 0.0602, 0.2503], device='cuda:1'), in_proj_covar=tensor([0.0208, 0.0271, 0.0341, 0.0166, 0.0203, 0.0193, 0.0252, 0.0351], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 08:58:27,113 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 08:58:45,517 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.872e+02 3.739e+02 4.989e+02 1.887e+03, threshold=7.477e+02, percent-clipped=4.0 2022-12-07 08:58:48,263 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:58:51,213 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.09 vs. limit=2.0 2022-12-07 08:58:55,686 INFO [train.py:873] (1/4) Epoch 3, batch 3200, loss[loss=0.1983, simple_loss=0.1929, pruned_loss=0.1018, over 5988.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2157, pruned_loss=0.1074, over 2018131.97 frames. ], batch size: 100, lr: 2.39e-02, grad_scale: 8.0 2022-12-07 08:59:22,797 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:59:48,463 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:03,917 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:07,367 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5859, 3.3977, 4.3958, 2.3697, 2.6598, 2.6226, 1.3913, 2.8737], device='cuda:1'), covar=tensor([0.1349, 0.0403, 0.0721, 0.2237, 0.1089, 0.1305, 0.3394, 0.1523], device='cuda:1'), in_proj_covar=tensor([0.0061, 0.0068, 0.0061, 0.0074, 0.0083, 0.0061, 0.0126, 0.0067], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:1') 2022-12-07 09:00:11,699 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 3.073e+02 3.846e+02 5.259e+02 1.192e+03, threshold=7.691e+02, percent-clipped=7.0 2022-12-07 09:00:22,200 INFO [train.py:873] (1/4) Epoch 3, batch 3300, loss[loss=0.2175, simple_loss=0.21, pruned_loss=0.1125, over 4957.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2153, pruned_loss=0.1069, over 2022381.57 frames. ], batch size: 100, lr: 2.38e-02, grad_scale: 8.0 2022-12-07 09:00:28,643 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.36 vs. limit=5.0 2022-12-07 09:00:29,905 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:31,635 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0244, 1.7387, 1.9805, 2.0175, 2.0729, 1.9868, 2.1481, 1.7522], device='cuda:1'), covar=tensor([0.0619, 0.1535, 0.0556, 0.0730, 0.0681, 0.0625, 0.0711, 0.0788], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0180, 0.0117, 0.0112, 0.0104, 0.0100, 0.0165, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:00:46,539 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.20 vs. limit=2.0 2022-12-07 09:00:56,521 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 09:01:16,390 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4996, 2.1525, 2.0224, 2.2020, 1.6707, 2.1622, 2.1661, 0.8600], device='cuda:1'), covar=tensor([0.3307, 0.1644, 0.3403, 0.1526, 0.1367, 0.0750, 0.1251, 0.4178], device='cuda:1'), in_proj_covar=tensor([0.0162, 0.0069, 0.0052, 0.0059, 0.0067, 0.0057, 0.0075, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:01:18,094 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:01:33,642 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:01:37,436 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 3.014e+02 3.926e+02 4.952e+02 1.093e+03, threshold=7.852e+02, percent-clipped=5.0 2022-12-07 09:01:48,040 INFO [train.py:873] (1/4) Epoch 3, batch 3400, loss[loss=0.2358, simple_loss=0.2, pruned_loss=0.1358, over 1168.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2148, pruned_loss=0.107, over 1933184.64 frames. ], batch size: 100, lr: 2.38e-02, grad_scale: 8.0 2022-12-07 09:01:52,023 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:01:53,142 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:02:10,088 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:02:25,475 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 09:02:33,763 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2497, 2.3085, 1.7729, 2.4684, 2.1847, 2.4170, 2.1231, 1.8254], device='cuda:1'), covar=tensor([0.0174, 0.0296, 0.0896, 0.0095, 0.0238, 0.0111, 0.0444, 0.0953], device='cuda:1'), in_proj_covar=tensor([0.0205, 0.0277, 0.0348, 0.0168, 0.0210, 0.0200, 0.0255, 0.0348], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:03:03,541 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 3.062e+02 4.109e+02 5.827e+02 9.337e+02, threshold=8.217e+02, percent-clipped=6.0 2022-12-07 09:03:06,419 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:03:11,927 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6827, 0.6412, 0.1519, 0.7150, 0.6424, 0.0854, 0.6486, 0.7440], device='cuda:1'), covar=tensor([0.0060, 0.0050, 0.0012, 0.0052, 0.0019, 0.0017, 0.0060, 0.0031], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0020, 0.0021, 0.0018, 0.0018, 0.0023, 0.0016, 0.0017], device='cuda:1'), out_proj_covar=tensor([4.8067e-05, 5.5937e-05, 5.4430e-05, 5.3496e-05, 4.9553e-05, 5.9092e-05, 4.8804e-05, 4.5520e-05], device='cuda:1') 2022-12-07 09:03:13,885 INFO [train.py:873] (1/4) Epoch 3, batch 3500, loss[loss=0.195, simple_loss=0.1754, pruned_loss=0.1073, over 2659.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.214, pruned_loss=0.1065, over 1919880.26 frames. ], batch size: 100, lr: 2.37e-02, grad_scale: 8.0 2022-12-07 09:03:28,624 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3656, 3.0391, 3.0572, 3.4385, 3.0509, 2.7027, 3.4425, 3.3750], device='cuda:1'), covar=tensor([0.0694, 0.0563, 0.0682, 0.0567, 0.0686, 0.0666, 0.0579, 0.0671], device='cuda:1'), in_proj_covar=tensor([0.0097, 0.0077, 0.0100, 0.0093, 0.0101, 0.0069, 0.0100, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:03:35,486 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0581, 0.8986, 0.6301, 0.9274, 0.9777, 0.6418, 1.2165, 1.2625], device='cuda:1'), covar=tensor([0.0351, 0.0302, 0.0155, 0.0536, 0.0249, 0.0188, 0.0149, 0.0161], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0020, 0.0021, 0.0019, 0.0018, 0.0023, 0.0017, 0.0017], device='cuda:1'), out_proj_covar=tensor([4.9272e-05, 5.6131e-05, 5.5417e-05, 5.6072e-05, 4.9805e-05, 5.9832e-05, 4.9787e-05, 4.5565e-05], device='cuda:1') 2022-12-07 09:03:39,415 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2022-12-07 09:03:46,488 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:03:54,768 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:04:28,294 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.925e+02 3.819e+02 4.873e+02 8.102e+02, threshold=7.639e+02, percent-clipped=0.0 2022-12-07 09:04:38,566 INFO [train.py:873] (1/4) Epoch 3, batch 3600, loss[loss=0.2111, simple_loss=0.2095, pruned_loss=0.1064, over 10311.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2143, pruned_loss=0.1063, over 1983172.29 frames. ], batch size: 100, lr: 2.37e-02, grad_scale: 8.0 2022-12-07 09:04:47,562 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:04:48,133 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 09:05:22,758 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 09:05:44,883 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2600, 1.0616, 1.1893, 1.6239, 1.0557, 0.7677, 1.5206, 0.8422], device='cuda:1'), covar=tensor([0.0439, 0.0407, 0.0610, 0.0319, 0.0568, 0.0184, 0.0173, 0.0404], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0012, 0.0012, 0.0014, 0.0013, 0.0015], device='cuda:1'), out_proj_covar=tensor([3.3294e-05, 3.2809e-05, 3.4269e-05, 3.3160e-05, 3.4860e-05, 3.6299e-05, 4.3171e-05, 4.4974e-05], device='cuda:1') 2022-12-07 09:05:54,915 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 3.054e+02 3.647e+02 5.491e+02 1.580e+03, threshold=7.295e+02, percent-clipped=4.0 2022-12-07 09:06:05,677 INFO [train.py:873] (1/4) Epoch 3, batch 3700, loss[loss=0.2246, simple_loss=0.2136, pruned_loss=0.1178, over 6919.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2144, pruned_loss=0.1068, over 1960661.81 frames. ], batch size: 100, lr: 2.36e-02, grad_scale: 8.0 2022-12-07 09:06:09,917 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:10,734 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:23,466 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:37,210 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6888, 1.4498, 2.8350, 1.6202, 2.8359, 2.8610, 2.0589, 2.9103], device='cuda:1'), covar=tensor([0.0170, 0.1878, 0.0185, 0.1530, 0.0204, 0.0227, 0.0632, 0.0158], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0145, 0.0104, 0.0157, 0.0120, 0.0111, 0.0099, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:06:39,734 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:06:40,671 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3328, 2.3888, 2.0008, 2.1017, 1.5899, 2.2119, 2.1273, 0.9616], device='cuda:1'), covar=tensor([0.3767, 0.1080, 0.1372, 0.1434, 0.1332, 0.0622, 0.1085, 0.3104], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0064, 0.0051, 0.0059, 0.0068, 0.0057, 0.0072, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:06:41,573 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5118, 1.9709, 2.7012, 2.2387, 2.8328, 2.3917, 2.4680, 2.1665], device='cuda:1'), covar=tensor([0.0132, 0.1035, 0.0135, 0.0548, 0.0170, 0.0241, 0.0227, 0.0913], device='cuda:1'), in_proj_covar=tensor([0.0215, 0.0335, 0.0186, 0.0283, 0.0229, 0.0224, 0.0242, 0.0352], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:06:44,383 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 09:06:51,534 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:52,327 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:07:21,106 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.825e+02 3.727e+02 4.870e+02 9.642e+02, threshold=7.453e+02, percent-clipped=3.0 2022-12-07 09:07:31,081 INFO [train.py:873] (1/4) Epoch 3, batch 3800, loss[loss=0.2026, simple_loss=0.179, pruned_loss=0.113, over 1230.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2143, pruned_loss=0.1067, over 1967031.98 frames. ], batch size: 100, lr: 2.35e-02, grad_scale: 8.0 2022-12-07 09:07:35,842 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:08:12,346 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-07 09:08:25,638 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 09:08:28,703 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:08:35,306 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2032, 3.1046, 3.9829, 2.6739, 3.0007, 3.1555, 1.3079, 2.9783], device='cuda:1'), covar=tensor([0.1539, 0.0525, 0.0595, 0.1184, 0.0844, 0.0871, 0.3306, 0.1118], device='cuda:1'), in_proj_covar=tensor([0.0059, 0.0066, 0.0061, 0.0072, 0.0085, 0.0063, 0.0127, 0.0069], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:08:47,290 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.617e+02 3.029e+02 3.748e+02 4.757e+02 1.336e+03, threshold=7.495e+02, percent-clipped=2.0 2022-12-07 09:08:58,126 INFO [train.py:873] (1/4) Epoch 3, batch 3900, loss[loss=0.2017, simple_loss=0.2095, pruned_loss=0.09692, over 14249.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2133, pruned_loss=0.1057, over 1942598.42 frames. ], batch size: 39, lr: 2.35e-02, grad_scale: 8.0 2022-12-07 09:09:02,337 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:09:15,782 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7076, 1.3738, 1.7155, 1.1861, 1.3166, 1.4810, 1.3963, 1.4737], device='cuda:1'), covar=tensor([0.0286, 0.1104, 0.0456, 0.0980, 0.0813, 0.0570, 0.0380, 0.0757], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0202, 0.0096, 0.0125, 0.0084, 0.0082, 0.0070, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:10:06,675 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.69 vs. limit=5.0 2022-12-07 09:10:13,854 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.994e+02 3.505e+02 4.615e+02 1.018e+03, threshold=7.009e+02, percent-clipped=3.0 2022-12-07 09:10:15,631 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7558, 2.8305, 2.3456, 1.4564, 2.5813, 2.5255, 2.9594, 2.0689], device='cuda:1'), covar=tensor([0.0419, 0.2636, 0.0995, 0.2467, 0.0938, 0.0356, 0.0505, 0.1261], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0203, 0.0096, 0.0124, 0.0085, 0.0081, 0.0068, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:10:23,710 INFO [train.py:873] (1/4) Epoch 3, batch 4000, loss[loss=0.2121, simple_loss=0.2168, pruned_loss=0.1038, over 13524.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2127, pruned_loss=0.1052, over 1940908.51 frames. ], batch size: 100, lr: 2.34e-02, grad_scale: 8.0 2022-12-07 09:10:27,285 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:10:42,447 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:10:53,302 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8012, 1.7283, 3.0025, 2.2181, 2.8775, 1.7143, 2.3445, 2.7547], device='cuda:1'), covar=tensor([0.0557, 0.3758, 0.0277, 0.4738, 0.0215, 0.3078, 0.1033, 0.0294], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0322, 0.0180, 0.0422, 0.0172, 0.0327, 0.0295, 0.0179], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:10:57,673 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:11:20,481 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:11:23,549 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:11:24,480 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7658, 2.8672, 2.4313, 1.5371, 2.6351, 2.5333, 2.8014, 2.1387], device='cuda:1'), covar=tensor([0.0331, 0.2284, 0.0852, 0.2625, 0.0613, 0.0324, 0.0525, 0.1115], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0205, 0.0096, 0.0126, 0.0085, 0.0083, 0.0068, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:11:35,433 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5253, 1.5551, 2.6650, 1.4085, 2.7348, 2.6783, 1.9205, 2.7843], device='cuda:1'), covar=tensor([0.0210, 0.1743, 0.0193, 0.1683, 0.0237, 0.0277, 0.0623, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0146, 0.0110, 0.0161, 0.0124, 0.0114, 0.0101, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:11:35,983 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 09:11:39,093 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:11:39,768 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.771e+02 3.795e+02 5.006e+02 9.210e+02, threshold=7.589e+02, percent-clipped=4.0 2022-12-07 09:11:45,095 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2022-12-07 09:11:50,340 INFO [train.py:873] (1/4) Epoch 3, batch 4100, loss[loss=0.2341, simple_loss=0.1994, pruned_loss=0.1344, over 1246.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2137, pruned_loss=0.1055, over 1999026.73 frames. ], batch size: 100, lr: 2.34e-02, grad_scale: 8.0 2022-12-07 09:12:19,201 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.99 vs. limit=2.0 2022-12-07 09:12:42,505 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:12:49,456 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4167, 3.1942, 3.2681, 3.3331, 3.2801, 2.9424, 1.2698, 3.1921], device='cuda:1'), covar=tensor([0.0227, 0.0386, 0.0677, 0.0309, 0.0410, 0.0462, 0.3249, 0.0340], device='cuda:1'), in_proj_covar=tensor([0.0110, 0.0117, 0.0112, 0.0091, 0.0149, 0.0102, 0.0148, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:13:05,131 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 2.865e+02 3.600e+02 4.882e+02 1.320e+03, threshold=7.199e+02, percent-clipped=6.0 2022-12-07 09:13:10,945 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2022-12-07 09:13:15,261 INFO [train.py:873] (1/4) Epoch 3, batch 4200, loss[loss=0.1773, simple_loss=0.1966, pruned_loss=0.07898, over 14266.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2149, pruned_loss=0.1067, over 1952265.03 frames. ], batch size: 60, lr: 2.33e-02, grad_scale: 8.0 2022-12-07 09:13:19,745 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:14:00,926 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3370, 3.0685, 2.2008, 3.4668, 3.0986, 3.4209, 2.6184, 2.2583], device='cuda:1'), covar=tensor([0.0248, 0.0620, 0.2526, 0.0115, 0.0286, 0.0326, 0.0768, 0.2911], device='cuda:1'), in_proj_covar=tensor([0.0207, 0.0279, 0.0348, 0.0169, 0.0217, 0.0205, 0.0254, 0.0344], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:14:01,562 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:14:12,852 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3981, 3.0312, 2.1404, 3.3977, 3.1141, 3.4499, 2.4367, 2.2881], device='cuda:1'), covar=tensor([0.0205, 0.0756, 0.2464, 0.0105, 0.0336, 0.0387, 0.0935, 0.3114], device='cuda:1'), in_proj_covar=tensor([0.0209, 0.0284, 0.0354, 0.0171, 0.0220, 0.0208, 0.0259, 0.0349], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:14:15,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:14:25,608 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0958, 0.6088, 1.0425, 0.8889, 0.8710, 0.4884, 1.3658, 1.5119], device='cuda:1'), covar=tensor([0.0574, 0.0463, 0.0141, 0.0882, 0.0843, 0.0225, 0.0516, 0.0227], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0020, 0.0020, 0.0017, 0.0018, 0.0022, 0.0016, 0.0017], device='cuda:1'), out_proj_covar=tensor([4.9527e-05, 5.6090e-05, 5.2009e-05, 5.2130e-05, 5.1160e-05, 5.9394e-05, 4.9626e-05, 4.7923e-05], device='cuda:1') 2022-12-07 09:14:31,170 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 3.180e+02 3.832e+02 4.985e+02 8.220e+02, threshold=7.663e+02, percent-clipped=2.0 2022-12-07 09:14:42,547 INFO [train.py:873] (1/4) Epoch 3, batch 4300, loss[loss=0.182, simple_loss=0.2059, pruned_loss=0.07905, over 14002.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2143, pruned_loss=0.1063, over 1990794.89 frames. ], batch size: 26, lr: 2.33e-02, grad_scale: 8.0 2022-12-07 09:15:08,480 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:15:34,012 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:15:57,829 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.698e+02 3.428e+02 4.598e+02 7.579e+02, threshold=6.855e+02, percent-clipped=0.0 2022-12-07 09:16:06,219 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 09:16:08,044 INFO [train.py:873] (1/4) Epoch 3, batch 4400, loss[loss=0.1914, simple_loss=0.2103, pruned_loss=0.08619, over 14281.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2141, pruned_loss=0.1047, over 2086518.57 frames. ], batch size: 44, lr: 2.32e-02, grad_scale: 8.0 2022-12-07 09:16:10,828 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-07 09:16:13,744 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6031, 2.5041, 2.0651, 1.9841, 2.5828, 2.5931, 2.6060, 2.5209], device='cuda:1'), covar=tensor([0.1476, 0.1230, 0.3243, 0.4601, 0.1016, 0.1223, 0.1825, 0.1639], device='cuda:1'), in_proj_covar=tensor([0.0217, 0.0183, 0.0259, 0.0327, 0.0204, 0.0235, 0.0248, 0.0204], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:17:00,305 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:23,968 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 2.902e+02 3.784e+02 4.919e+02 1.720e+03, threshold=7.568e+02, percent-clipped=8.0 2022-12-07 09:17:28,323 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:34,597 INFO [train.py:873] (1/4) Epoch 3, batch 4500, loss[loss=0.1892, simple_loss=0.1672, pruned_loss=0.1056, over 2600.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2135, pruned_loss=0.1041, over 2015255.69 frames. ], batch size: 100, lr: 2.31e-02, grad_scale: 8.0 2022-12-07 09:17:38,130 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2069, 1.3484, 4.0162, 1.8936, 3.9748, 3.8845, 3.2909, 4.4595], device='cuda:1'), covar=tensor([0.0155, 0.2875, 0.0253, 0.2184, 0.0259, 0.0308, 0.0490, 0.0142], device='cuda:1'), in_proj_covar=tensor([0.0120, 0.0145, 0.0109, 0.0160, 0.0124, 0.0117, 0.0101, 0.0101], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:17:42,286 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:56,633 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6890, 3.3962, 4.6404, 3.3397, 4.4326, 4.4510, 3.9206, 3.9030], device='cuda:1'), covar=tensor([0.0057, 0.0887, 0.0061, 0.0436, 0.0150, 0.0127, 0.0660, 0.0617], device='cuda:1'), in_proj_covar=tensor([0.0212, 0.0333, 0.0201, 0.0286, 0.0240, 0.0226, 0.0251, 0.0360], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 09:18:20,846 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:18:50,393 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.725e+01 3.027e+02 3.828e+02 4.881e+02 1.178e+03, threshold=7.656e+02, percent-clipped=6.0 2022-12-07 09:18:59,480 INFO [train.py:873] (1/4) Epoch 3, batch 4600, loss[loss=0.2011, simple_loss=0.2128, pruned_loss=0.09464, over 14278.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.214, pruned_loss=0.1043, over 2048491.74 frames. ], batch size: 76, lr: 2.31e-02, grad_scale: 8.0 2022-12-07 09:19:00,429 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3222, 1.3902, 1.5853, 1.6125, 1.3662, 1.4969, 1.3507, 0.8973], device='cuda:1'), covar=tensor([0.2536, 0.1159, 0.0717, 0.0487, 0.1654, 0.0456, 0.1372, 0.2573], device='cuda:1'), in_proj_covar=tensor([0.0161, 0.0066, 0.0052, 0.0059, 0.0069, 0.0060, 0.0076, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:19:21,718 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:19:52,088 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:19:54,575 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7912, 1.9310, 3.1061, 3.0378, 2.9877, 2.1346, 2.9809, 2.2348], device='cuda:1'), covar=tensor([0.0055, 0.0138, 0.0110, 0.0080, 0.0042, 0.0223, 0.0029, 0.0189], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0158, 0.0198, 0.0168, 0.0142, 0.0202, 0.0106, 0.0195], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:1') 2022-12-07 09:20:16,184 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 3.281e+02 3.991e+02 5.326e+02 1.100e+03, threshold=7.982e+02, percent-clipped=6.0 2022-12-07 09:20:26,813 INFO [train.py:873] (1/4) Epoch 3, batch 4700, loss[loss=0.2035, simple_loss=0.2213, pruned_loss=0.09288, over 14242.00 frames. ], tot_loss[loss=0.212, simple_loss=0.214, pruned_loss=0.105, over 1969450.81 frames. ], batch size: 35, lr: 2.30e-02, grad_scale: 8.0 2022-12-07 09:20:33,476 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:21:43,695 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 3.110e+02 3.871e+02 4.666e+02 7.831e+02, threshold=7.741e+02, percent-clipped=0.0 2022-12-07 09:21:47,374 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2937, 1.4001, 1.1119, 1.2603, 1.4706, 0.5517, 1.4508, 1.1254], device='cuda:1'), covar=tensor([0.0655, 0.0530, 0.1567, 0.0506, 0.0418, 0.0257, 0.0468, 0.0511], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0012, 0.0011, 0.0012, 0.0014, 0.0013, 0.0016], device='cuda:1'), out_proj_covar=tensor([3.7679e-05, 3.5426e-05, 3.9325e-05, 3.4823e-05, 3.7078e-05, 3.6895e-05, 4.7652e-05, 4.9215e-05], device='cuda:1') 2022-12-07 09:21:52,844 INFO [train.py:873] (1/4) Epoch 3, batch 4800, loss[loss=0.2073, simple_loss=0.2137, pruned_loss=0.1005, over 14256.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2128, pruned_loss=0.1035, over 2026732.04 frames. ], batch size: 76, lr: 2.30e-02, grad_scale: 8.0 2022-12-07 09:22:12,728 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2022-12-07 09:22:35,286 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:23:13,338 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.781e+02 3.610e+02 5.418e+02 1.055e+03, threshold=7.220e+02, percent-clipped=5.0 2022-12-07 09:23:19,283 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:23:23,446 INFO [train.py:873] (1/4) Epoch 3, batch 4900, loss[loss=0.2177, simple_loss=0.216, pruned_loss=0.1096, over 14533.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2122, pruned_loss=0.1037, over 2005476.48 frames. ], batch size: 49, lr: 2.29e-02, grad_scale: 8.0 2022-12-07 09:23:44,378 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:10,881 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:17,518 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:25,506 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:37,074 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1324, 2.9737, 3.0625, 3.0533, 3.0240, 2.4973, 1.1902, 2.9184], device='cuda:1'), covar=tensor([0.0239, 0.0429, 0.0529, 0.0467, 0.0423, 0.0955, 0.3232, 0.0336], device='cuda:1'), in_proj_covar=tensor([0.0109, 0.0114, 0.0109, 0.0090, 0.0150, 0.0101, 0.0147, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:24:37,987 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:39,472 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 3.078e+02 3.886e+02 5.087e+02 8.123e+02, threshold=7.771e+02, percent-clipped=4.0 2022-12-07 09:24:48,396 INFO [train.py:873] (1/4) Epoch 3, batch 5000, loss[loss=0.2363, simple_loss=0.2218, pruned_loss=0.1254, over 7742.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2126, pruned_loss=0.1044, over 1930702.60 frames. ], batch size: 100, lr: 2.29e-02, grad_scale: 8.0 2022-12-07 09:25:10,511 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:25:20,676 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9159, 1.7346, 1.4670, 1.7444, 1.4263, 1.7447, 1.8082, 0.7676], device='cuda:1'), covar=tensor([0.4392, 0.1763, 0.2907, 0.1301, 0.1741, 0.1482, 0.1749, 0.4297], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0065, 0.0052, 0.0057, 0.0069, 0.0058, 0.0076, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:25:29,545 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:25:56,545 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:26:04,801 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 2.928e+02 3.762e+02 4.834e+02 9.442e+02, threshold=7.523e+02, percent-clipped=4.0 2022-12-07 09:26:07,841 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7281, 1.2184, 2.0161, 1.9370, 2.0634, 1.9843, 1.5212, 2.0094], device='cuda:1'), covar=tensor([0.0342, 0.0694, 0.0092, 0.0158, 0.0119, 0.0084, 0.0226, 0.0109], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0144, 0.0084, 0.0115, 0.0094, 0.0102, 0.0078, 0.0082], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:26:14,945 INFO [train.py:873] (1/4) Epoch 3, batch 5100, loss[loss=0.2007, simple_loss=0.2117, pruned_loss=0.09481, over 14278.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2116, pruned_loss=0.1031, over 1992512.24 frames. ], batch size: 31, lr: 2.28e-02, grad_scale: 8.0 2022-12-07 09:26:49,104 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:26:56,578 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:27:26,322 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 09:27:30,743 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.933e+02 3.750e+02 4.548e+02 8.499e+02, threshold=7.500e+02, percent-clipped=1.0 2022-12-07 09:27:37,610 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:27:39,976 INFO [train.py:873] (1/4) Epoch 3, batch 5200, loss[loss=0.2155, simple_loss=0.2225, pruned_loss=0.1042, over 14457.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2123, pruned_loss=0.1036, over 1973690.81 frames. ], batch size: 51, lr: 2.28e-02, grad_scale: 8.0 2022-12-07 09:28:01,610 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2014, 4.0176, 4.4196, 3.4817, 4.1461, 4.3104, 1.6612, 3.9623], device='cuda:1'), covar=tensor([0.0151, 0.0241, 0.0280, 0.0440, 0.0249, 0.0183, 0.2946, 0.0226], device='cuda:1'), in_proj_covar=tensor([0.0111, 0.0116, 0.0111, 0.0092, 0.0153, 0.0102, 0.0149, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:28:23,115 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:28:34,166 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:28:56,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.824e+02 3.945e+02 4.935e+02 1.895e+03, threshold=7.890e+02, percent-clipped=6.0 2022-12-07 09:28:59,429 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:06,135 INFO [train.py:873] (1/4) Epoch 3, batch 5300, loss[loss=0.212, simple_loss=0.2099, pruned_loss=0.107, over 14226.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2131, pruned_loss=0.1037, over 2015226.03 frames. ], batch size: 89, lr: 2.27e-02, grad_scale: 4.0 2022-12-07 09:29:23,378 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:26,889 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:43,756 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:52,914 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:30:07,590 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 09:30:23,816 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 2.763e+02 3.648e+02 4.928e+02 1.065e+03, threshold=7.297e+02, percent-clipped=2.0 2022-12-07 09:30:32,156 INFO [train.py:873] (1/4) Epoch 3, batch 5400, loss[loss=0.2145, simple_loss=0.2009, pruned_loss=0.1141, over 3876.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2123, pruned_loss=0.1031, over 2011580.77 frames. ], batch size: 100, lr: 2.27e-02, grad_scale: 4.0 2022-12-07 09:30:40,784 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3783, 4.2957, 3.8721, 3.9279, 4.0433, 4.2522, 4.4169, 4.3391], device='cuda:1'), covar=tensor([0.0686, 0.0542, 0.1699, 0.2052, 0.0673, 0.0481, 0.0830, 0.0835], device='cuda:1'), in_proj_covar=tensor([0.0219, 0.0191, 0.0272, 0.0343, 0.0217, 0.0247, 0.0260, 0.0215], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:31:00,063 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.33 vs. limit=2.0 2022-12-07 09:31:02,937 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:31:27,537 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:31:50,018 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.948e+02 3.775e+02 4.848e+02 6.830e+02, threshold=7.550e+02, percent-clipped=0.0 2022-12-07 09:31:59,001 INFO [train.py:873] (1/4) Epoch 3, batch 5500, loss[loss=0.2069, simple_loss=0.2174, pruned_loss=0.0982, over 14224.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2108, pruned_loss=0.1018, over 2032022.42 frames. ], batch size: 57, lr: 2.26e-02, grad_scale: 4.0 2022-12-07 09:32:19,614 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:32:26,484 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-07 09:32:31,233 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7230, 1.5244, 2.8988, 1.9779, 2.7497, 1.5683, 2.1204, 2.7301], device='cuda:1'), covar=tensor([0.0491, 0.4555, 0.0360, 0.7326, 0.0301, 0.3872, 0.1424, 0.0287], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0310, 0.0183, 0.0418, 0.0182, 0.0326, 0.0295, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:32:42,430 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:32:51,516 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1400, 0.8768, 1.0080, 1.0232, 0.9437, 0.6348, 1.1604, 1.0719], device='cuda:1'), covar=tensor([0.0302, 0.0369, 0.0137, 0.0234, 0.0360, 0.0191, 0.0393, 0.0420], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0019, 0.0019, 0.0016, 0.0017, 0.0022, 0.0017, 0.0017], device='cuda:1'), out_proj_covar=tensor([4.9517e-05, 5.4694e-05, 5.1248e-05, 5.1711e-05, 5.0355e-05, 6.1228e-05, 5.3387e-05, 4.9747e-05], device='cuda:1') 2022-12-07 09:33:15,209 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 09:33:15,378 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.742e+02 3.691e+02 4.963e+02 1.196e+03, threshold=7.382e+02, percent-clipped=6.0 2022-12-07 09:33:22,772 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:23,570 INFO [train.py:873] (1/4) Epoch 3, batch 5600, loss[loss=0.2533, simple_loss=0.2364, pruned_loss=0.1351, over 10338.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2125, pruned_loss=0.1041, over 2018481.14 frames. ], batch size: 100, lr: 2.26e-02, grad_scale: 8.0 2022-12-07 09:33:32,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 09:33:40,096 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:41,284 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:00,951 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:06,487 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:11,654 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8015, 2.3597, 2.3216, 1.1971, 2.2673, 2.5033, 2.4664, 1.9141], device='cuda:1'), covar=tensor([0.0578, 0.5577, 0.1446, 0.4384, 0.1372, 0.0796, 0.2035, 0.2237], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0226, 0.0103, 0.0133, 0.0089, 0.0089, 0.0074, 0.0109], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:34:24,036 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:37,259 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 09:34:43,762 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 3.037e+02 4.375e+02 5.410e+02 1.130e+03, threshold=8.749e+02, percent-clipped=7.0 2022-12-07 09:34:44,738 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:53,314 INFO [train.py:873] (1/4) Epoch 3, batch 5700, loss[loss=0.2124, simple_loss=0.2091, pruned_loss=0.1078, over 7783.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2114, pruned_loss=0.1036, over 1967315.26 frames. ], batch size: 100, lr: 2.25e-02, grad_scale: 8.0 2022-12-07 09:34:55,353 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 09:35:24,803 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:35:37,716 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6019, 1.9594, 3.9745, 3.8459, 3.9978, 2.5127, 3.8254, 2.6654], device='cuda:1'), covar=tensor([0.0056, 0.0193, 0.0147, 0.0075, 0.0034, 0.0292, 0.0024, 0.0241], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0159, 0.0197, 0.0171, 0.0142, 0.0206, 0.0109, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:1') 2022-12-07 09:36:06,592 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:36:09,992 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 09:36:12,609 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.426e+01 2.770e+02 3.515e+02 4.337e+02 1.206e+03, threshold=7.030e+02, percent-clipped=1.0 2022-12-07 09:36:21,397 INFO [train.py:873] (1/4) Epoch 3, batch 5800, loss[loss=0.1708, simple_loss=0.1979, pruned_loss=0.07183, over 14098.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2119, pruned_loss=0.1043, over 1921010.01 frames. ], batch size: 29, lr: 2.25e-02, grad_scale: 8.0 2022-12-07 09:36:37,676 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:36:42,575 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3581, 5.2706, 4.9544, 5.5389, 5.1706, 4.3966, 5.5817, 5.4053], device='cuda:1'), covar=tensor([0.0739, 0.0440, 0.0636, 0.0570, 0.0759, 0.0574, 0.0645, 0.0806], device='cuda:1'), in_proj_covar=tensor([0.0099, 0.0083, 0.0100, 0.0099, 0.0110, 0.0074, 0.0105, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:36:58,299 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-07 09:37:39,551 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.882e+02 3.467e+02 4.356e+02 1.149e+03, threshold=6.934e+02, percent-clipped=1.0 2022-12-07 09:37:47,930 INFO [train.py:873] (1/4) Epoch 3, batch 5900, loss[loss=0.2192, simple_loss=0.2157, pruned_loss=0.1114, over 9494.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2111, pruned_loss=0.1023, over 1971724.17 frames. ], batch size: 100, lr: 2.24e-02, grad_scale: 8.0 2022-12-07 09:38:04,843 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:11,947 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:32,095 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:47,677 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:03,326 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 09:39:07,989 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:08,235 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 09:39:09,614 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 3.189e+02 4.370e+02 6.428e+02 1.112e+03, threshold=8.740e+02, percent-clipped=21.0 2022-12-07 09:39:12,494 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:16,015 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:18,491 INFO [train.py:873] (1/4) Epoch 3, batch 6000, loss[loss=0.2177, simple_loss=0.2273, pruned_loss=0.1041, over 14677.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2118, pruned_loss=0.1037, over 1966543.17 frames. ], batch size: 33, lr: 2.24e-02, grad_scale: 8.0 2022-12-07 09:39:18,492 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 09:39:27,744 INFO [train.py:905] (1/4) Epoch 3, validation: loss=0.1295, simple_loss=0.172, pruned_loss=0.04354, over 857387.00 frames. 2022-12-07 09:39:27,747 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 09:39:38,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9480, 3.4987, 3.5996, 4.0366, 3.6599, 3.1874, 4.1069, 4.0112], device='cuda:1'), covar=tensor([0.0704, 0.0714, 0.0631, 0.0490, 0.0663, 0.0523, 0.0519, 0.0623], device='cuda:1'), in_proj_covar=tensor([0.0095, 0.0078, 0.0094, 0.0093, 0.0103, 0.0068, 0.0099, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:40:01,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0605, 0.6108, 1.2323, 0.7824, 0.9377, 0.6549, 1.5048, 1.2478], device='cuda:1'), covar=tensor([0.0764, 0.0575, 0.0232, 0.1173, 0.0630, 0.0197, 0.0419, 0.0227], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0021, 0.0020, 0.0017, 0.0017, 0.0023, 0.0018, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.5068e-05, 6.1358e-05, 5.4723e-05, 5.4665e-05, 5.2508e-05, 6.4742e-05, 5.8764e-05, 5.1517e-05], device='cuda:1') 2022-12-07 09:40:01,538 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.60 vs. limit=5.0 2022-12-07 09:40:02,968 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3839, 4.1435, 3.8816, 3.9311, 4.1309, 4.1977, 4.4311, 4.3826], device='cuda:1'), covar=tensor([0.0766, 0.0478, 0.1484, 0.2436, 0.0621, 0.0596, 0.0851, 0.0858], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0195, 0.0268, 0.0354, 0.0220, 0.0251, 0.0262, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:40:16,072 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:40:19,039 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 09:40:47,317 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.723e+02 3.536e+02 4.487e+02 9.124e+02, threshold=7.073e+02, percent-clipped=3.0 2022-12-07 09:40:56,102 INFO [train.py:873] (1/4) Epoch 3, batch 6100, loss[loss=0.2364, simple_loss=0.2302, pruned_loss=0.1213, over 14509.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2117, pruned_loss=0.1035, over 1956481.31 frames. ], batch size: 34, lr: 2.23e-02, grad_scale: 8.0 2022-12-07 09:41:01,726 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2022-12-07 09:41:12,168 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:41:35,265 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5638, 1.4819, 4.1431, 2.0388, 4.1901, 4.2193, 3.9080, 4.7151], device='cuda:1'), covar=tensor([0.0122, 0.2551, 0.0263, 0.1844, 0.0280, 0.0226, 0.0221, 0.0108], device='cuda:1'), in_proj_covar=tensor([0.0124, 0.0149, 0.0112, 0.0158, 0.0130, 0.0117, 0.0101, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:41:47,266 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3177, 1.5520, 2.6172, 2.0821, 2.3406, 1.6799, 2.1754, 2.3169], device='cuda:1'), covar=tensor([0.0527, 0.2572, 0.0244, 0.2917, 0.0149, 0.2090, 0.0853, 0.0319], device='cuda:1'), in_proj_covar=tensor([0.0242, 0.0301, 0.0173, 0.0407, 0.0177, 0.0316, 0.0284, 0.0176], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:41:54,625 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:11,603 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:12,949 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2022-12-07 09:42:14,077 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 3.459e+02 4.303e+02 5.209e+02 1.109e+03, threshold=8.606e+02, percent-clipped=4.0 2022-12-07 09:42:22,734 INFO [train.py:873] (1/4) Epoch 3, batch 6200, loss[loss=0.2182, simple_loss=0.2279, pruned_loss=0.1043, over 14023.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.211, pruned_loss=0.103, over 1920598.85 frames. ], batch size: 22, lr: 2.23e-02, grad_scale: 8.0 2022-12-07 09:42:25,278 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:39,395 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 09:43:04,560 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:11,367 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 09:43:18,511 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:21,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9199, 3.7232, 3.6126, 4.0521, 3.6535, 3.2781, 4.0250, 3.9776], device='cuda:1'), covar=tensor([0.0807, 0.0524, 0.0642, 0.0611, 0.0754, 0.0495, 0.0721, 0.0670], device='cuda:1'), in_proj_covar=tensor([0.0101, 0.0082, 0.0100, 0.0097, 0.0107, 0.0072, 0.0104, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:43:34,696 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:41,036 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.777e+02 3.357e+02 4.403e+02 6.655e+02, threshold=6.715e+02, percent-clipped=0.0 2022-12-07 09:43:49,957 INFO [train.py:873] (1/4) Epoch 3, batch 6300, loss[loss=0.2147, simple_loss=0.2187, pruned_loss=0.1053, over 13524.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2108, pruned_loss=0.1024, over 1923708.48 frames. ], batch size: 100, lr: 2.22e-02, grad_scale: 8.0 2022-12-07 09:44:08,371 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.80 vs. limit=5.0 2022-12-07 09:44:32,707 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:44:47,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8699, 1.9320, 3.9529, 4.1067, 4.1338, 2.5033, 4.1949, 3.1197], device='cuda:1'), covar=tensor([0.0049, 0.0190, 0.0170, 0.0059, 0.0025, 0.0270, 0.0026, 0.0195], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0161, 0.0204, 0.0174, 0.0146, 0.0208, 0.0113, 0.0201], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:45:07,700 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.996e+02 3.929e+02 5.404e+02 9.451e+02, threshold=7.857e+02, percent-clipped=12.0 2022-12-07 09:45:16,311 INFO [train.py:873] (1/4) Epoch 3, batch 6400, loss[loss=0.2069, simple_loss=0.1776, pruned_loss=0.1181, over 2655.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2095, pruned_loss=0.1008, over 1969791.98 frames. ], batch size: 100, lr: 2.22e-02, grad_scale: 8.0 2022-12-07 09:45:19,375 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-07 09:45:35,954 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1762, 2.9438, 2.7077, 2.7449, 3.0738, 3.0099, 3.2188, 3.1260], device='cuda:1'), covar=tensor([0.0915, 0.0922, 0.1876, 0.3362, 0.0803, 0.0840, 0.0939, 0.0992], device='cuda:1'), in_proj_covar=tensor([0.0220, 0.0191, 0.0270, 0.0354, 0.0218, 0.0249, 0.0260, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:46:05,545 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7515, 1.2712, 2.4265, 2.3870, 2.3684, 2.3841, 2.0050, 2.4583], device='cuda:1'), covar=tensor([0.0503, 0.0799, 0.0084, 0.0164, 0.0176, 0.0097, 0.0202, 0.0109], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0148, 0.0085, 0.0120, 0.0099, 0.0106, 0.0081, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:46:16,482 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:46:32,460 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:46:34,034 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.688e+02 3.501e+02 4.661e+02 7.945e+02, threshold=7.001e+02, percent-clipped=1.0 2022-12-07 09:46:43,052 INFO [train.py:873] (1/4) Epoch 3, batch 6500, loss[loss=0.1801, simple_loss=0.1943, pruned_loss=0.08295, over 13941.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2096, pruned_loss=0.1009, over 1986063.76 frames. ], batch size: 23, lr: 2.21e-02, grad_scale: 8.0 2022-12-07 09:46:58,411 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0670, 1.7501, 1.8050, 1.0091, 1.6164, 1.8816, 2.1287, 1.6181], device='cuda:1'), covar=tensor([0.0521, 0.2193, 0.1319, 0.3760, 0.0999, 0.0629, 0.0500, 0.1776], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0223, 0.0105, 0.0130, 0.0089, 0.0085, 0.0078, 0.0110], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:47:09,443 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:15,497 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0463, 1.7936, 1.6040, 2.0766, 1.8047, 2.0666, 1.8591, 1.7981], device='cuda:1'), covar=tensor([0.0132, 0.0358, 0.0493, 0.0091, 0.0216, 0.0072, 0.0305, 0.0261], device='cuda:1'), in_proj_covar=tensor([0.0214, 0.0282, 0.0326, 0.0173, 0.0223, 0.0217, 0.0262, 0.0336], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:47:19,602 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:24,878 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:32,437 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0311, 1.9825, 1.7289, 1.8398, 1.3309, 1.7426, 1.7630, 0.8058], device='cuda:1'), covar=tensor([0.3526, 0.1022, 0.2649, 0.0716, 0.1248, 0.0776, 0.2202, 0.3465], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0062, 0.0051, 0.0057, 0.0067, 0.0057, 0.0077, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:47:33,136 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:54,480 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:00,153 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 3.136e+02 4.193e+02 5.448e+02 1.293e+03, threshold=8.387e+02, percent-clipped=8.0 2022-12-07 09:48:08,231 INFO [train.py:873] (1/4) Epoch 3, batch 6600, loss[loss=0.2362, simple_loss=0.1943, pruned_loss=0.139, over 1224.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2109, pruned_loss=0.1019, over 1990824.43 frames. ], batch size: 100, lr: 2.21e-02, grad_scale: 8.0 2022-12-07 09:48:32,384 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3161, 2.1823, 2.0232, 1.2529, 1.9947, 2.2134, 2.2305, 1.8745], device='cuda:1'), covar=tensor([0.0707, 0.2872, 0.1230, 0.3044, 0.1016, 0.0633, 0.0801, 0.1414], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0216, 0.0102, 0.0126, 0.0088, 0.0085, 0.0075, 0.0107], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:48:34,765 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:40,823 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:51,210 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:10,315 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=16.99 vs. limit=5.0 2022-12-07 09:49:20,337 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0383, 2.6660, 1.9750, 3.1578, 2.8231, 3.0590, 2.5488, 2.1178], device='cuda:1'), covar=tensor([0.0337, 0.0961, 0.3194, 0.0148, 0.0464, 0.0531, 0.1010, 0.3667], device='cuda:1'), in_proj_covar=tensor([0.0211, 0.0279, 0.0324, 0.0173, 0.0223, 0.0212, 0.0256, 0.0326], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:49:27,684 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.849e+02 3.705e+02 5.166e+02 8.548e+02, threshold=7.409e+02, percent-clipped=1.0 2022-12-07 09:49:33,534 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9008, 1.5902, 4.0629, 3.8629, 3.8984, 4.0161, 3.6712, 4.0865], device='cuda:1'), covar=tensor([0.1223, 0.1339, 0.0095, 0.0127, 0.0123, 0.0103, 0.0167, 0.0119], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0148, 0.0085, 0.0118, 0.0098, 0.0104, 0.0080, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:49:34,275 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:36,121 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:36,764 INFO [train.py:873] (1/4) Epoch 3, batch 6700, loss[loss=0.2017, simple_loss=0.2026, pruned_loss=0.1005, over 13881.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2103, pruned_loss=0.1007, over 2061093.96 frames. ], batch size: 23, lr: 2.20e-02, grad_scale: 8.0 2022-12-07 09:49:57,717 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:50:14,039 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8441, 1.3006, 3.4326, 3.2868, 3.3842, 3.3844, 2.8153, 3.4635], device='cuda:1'), covar=tensor([0.1193, 0.1425, 0.0123, 0.0183, 0.0163, 0.0126, 0.0241, 0.0145], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0148, 0.0085, 0.0118, 0.0098, 0.0104, 0.0080, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:50:51,717 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:50:54,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.679e+02 3.700e+02 5.034e+02 8.925e+02, threshold=7.401e+02, percent-clipped=4.0 2022-12-07 09:51:03,449 INFO [train.py:873] (1/4) Epoch 3, batch 6800, loss[loss=0.1964, simple_loss=0.1786, pruned_loss=0.107, over 2634.00 frames. ], tot_loss[loss=0.206, simple_loss=0.21, pruned_loss=0.101, over 2027391.10 frames. ], batch size: 100, lr: 2.20e-02, grad_scale: 8.0 2022-12-07 09:51:26,045 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:36,891 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0841, 1.0827, 0.7685, 1.0343, 1.0738, 0.5810, 1.0969, 0.8520], device='cuda:1'), covar=tensor([0.0263, 0.0198, 0.0117, 0.0354, 0.0298, 0.0153, 0.0320, 0.0228], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0018, 0.0019, 0.0015, 0.0016, 0.0022, 0.0018, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.2617e-05, 5.4746e-05, 5.2287e-05, 5.2049e-05, 4.9464e-05, 6.1972e-05, 5.8360e-05, 5.0859e-05], device='cuda:1') 2022-12-07 09:51:40,276 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:41,076 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:42,002 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:54,846 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:00,700 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 09:52:21,335 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.962e+02 3.808e+02 5.028e+02 7.344e+02, threshold=7.616e+02, percent-clipped=0.0 2022-12-07 09:52:22,063 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2022-12-07 09:52:22,243 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:23,212 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2430, 2.7049, 3.5587, 2.2833, 2.4691, 2.9558, 1.3958, 2.7422], device='cuda:1'), covar=tensor([0.1550, 0.0784, 0.0885, 0.1891, 0.1672, 0.0846, 0.4998, 0.1233], device='cuda:1'), in_proj_covar=tensor([0.0062, 0.0066, 0.0069, 0.0074, 0.0087, 0.0060, 0.0135, 0.0068], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:52:25,196 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=4.16 vs. limit=2.0 2022-12-07 09:52:25,647 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3552, 3.9603, 3.8219, 4.2925, 4.2840, 3.9200, 4.3311, 3.6861], device='cuda:1'), covar=tensor([0.0342, 0.0829, 0.0359, 0.0494, 0.0514, 0.0642, 0.0560, 0.0504], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0192, 0.0124, 0.0116, 0.0121, 0.0103, 0.0180, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:52:29,726 INFO [train.py:873] (1/4) Epoch 3, batch 6900, loss[loss=0.1489, simple_loss=0.1664, pruned_loss=0.06567, over 9986.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2093, pruned_loss=0.1003, over 2004766.78 frames. ], batch size: 12, lr: 2.19e-02, grad_scale: 8.0 2022-12-07 09:52:34,983 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:35,748 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:45,118 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7626, 1.2984, 1.5508, 1.0345, 1.2320, 1.4695, 1.4913, 1.3330], device='cuda:1'), covar=tensor([0.0330, 0.1030, 0.0633, 0.1454, 0.0947, 0.0872, 0.0471, 0.1264], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0228, 0.0104, 0.0126, 0.0090, 0.0086, 0.0076, 0.0112], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:52:48,003 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-07 09:53:46,969 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.783e+02 2.990e+02 4.011e+02 5.296e+02 1.371e+03, threshold=8.022e+02, percent-clipped=6.0 2022-12-07 09:53:50,229 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:53:55,140 INFO [train.py:873] (1/4) Epoch 3, batch 7000, loss[loss=0.2128, simple_loss=0.2114, pruned_loss=0.1071, over 14256.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2095, pruned_loss=0.1011, over 1968475.17 frames. ], batch size: 60, lr: 2.19e-02, grad_scale: 8.0 2022-12-07 09:53:59,784 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:54:03,589 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.24 vs. limit=2.0 2022-12-07 09:54:39,489 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:54:52,519 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:05,119 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:07,853 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7873, 3.2633, 4.2166, 2.8414, 2.5086, 2.9498, 1.5338, 3.0596], device='cuda:1'), covar=tensor([0.1879, 0.0621, 0.0892, 0.1969, 0.1766, 0.1034, 0.5538, 0.1371], device='cuda:1'), in_proj_covar=tensor([0.0060, 0.0068, 0.0069, 0.0074, 0.0086, 0.0061, 0.0135, 0.0067], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:55:12,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.681e+02 3.418e+02 4.673e+02 9.486e+02, threshold=6.836e+02, percent-clipped=4.0 2022-12-07 09:55:13,376 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:21,728 INFO [train.py:873] (1/4) Epoch 3, batch 7100, loss[loss=0.1946, simple_loss=0.2065, pruned_loss=0.09136, over 14220.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2087, pruned_loss=0.1003, over 1947311.43 frames. ], batch size: 80, lr: 2.18e-02, grad_scale: 8.0 2022-12-07 09:55:32,441 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:32,461 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:44,088 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:59,951 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:02,572 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8971, 4.5581, 4.4970, 4.9244, 4.7555, 4.2713, 4.9216, 4.1199], device='cuda:1'), covar=tensor([0.0307, 0.1023, 0.0259, 0.0350, 0.0516, 0.0493, 0.0447, 0.0468], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0188, 0.0123, 0.0114, 0.0120, 0.0100, 0.0176, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 09:56:05,977 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:06,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8097, 0.9698, 0.6573, 1.0066, 0.9887, 0.6191, 0.8694, 1.0014], device='cuda:1'), covar=tensor([0.0233, 0.0516, 0.0129, 0.0178, 0.0279, 0.0151, 0.0214, 0.0166], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0018, 0.0019, 0.0014, 0.0015, 0.0022, 0.0017, 0.0016], device='cuda:1'), out_proj_covar=tensor([5.4620e-05, 5.4919e-05, 5.3555e-05, 5.1006e-05, 4.8582e-05, 6.1700e-05, 5.8053e-05, 5.0019e-05], device='cuda:1') 2022-12-07 09:56:11,641 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2022-12-07 09:56:13,511 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2165, 2.9064, 2.2278, 3.3444, 3.0423, 3.2275, 2.7694, 2.1871], device='cuda:1'), covar=tensor([0.0373, 0.0791, 0.2801, 0.0202, 0.0490, 0.0484, 0.0766, 0.3369], device='cuda:1'), in_proj_covar=tensor([0.0214, 0.0282, 0.0329, 0.0179, 0.0228, 0.0212, 0.0259, 0.0327], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:1') 2022-12-07 09:56:24,901 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:25,597 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:33,509 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0638, 3.1373, 2.4201, 2.5080, 2.9378, 3.0749, 3.2099, 3.0424], device='cuda:1'), covar=tensor([0.1427, 0.1030, 0.3412, 0.4126, 0.1774, 0.1208, 0.1599, 0.1540], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0198, 0.0279, 0.0363, 0.0222, 0.0259, 0.0263, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:56:38,993 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.756e+01 2.795e+02 3.480e+02 4.240e+02 8.610e+02, threshold=6.960e+02, percent-clipped=4.0 2022-12-07 09:56:40,766 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:47,594 INFO [train.py:873] (1/4) Epoch 3, batch 7200, loss[loss=0.1879, simple_loss=0.2003, pruned_loss=0.08773, over 13871.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2098, pruned_loss=0.1019, over 1921273.69 frames. ], batch size: 23, lr: 2.18e-02, grad_scale: 8.0 2022-12-07 09:56:48,421 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:57:29,636 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2156, 1.0253, 0.9189, 1.3065, 1.0075, 0.9217, 1.0354, 1.2556], device='cuda:1'), covar=tensor([0.0677, 0.1723, 0.1553, 0.0667, 0.0583, 0.0387, 0.0296, 0.1012], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0011, 0.0011, 0.0011, 0.0011, 0.0014, 0.0013, 0.0016], device='cuda:1'), out_proj_covar=tensor([3.8688e-05, 3.6665e-05, 4.0395e-05, 3.7706e-05, 3.7227e-05, 4.3339e-05, 5.1530e-05, 5.2949e-05], device='cuda:1') 2022-12-07 09:57:39,290 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:05,274 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.951e+02 3.875e+02 4.878e+02 1.117e+03, threshold=7.750e+02, percent-clipped=7.0 2022-12-07 09:58:05,377 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6475, 3.4259, 3.4396, 3.8198, 3.4475, 2.8080, 3.7332, 3.7252], device='cuda:1'), covar=tensor([0.0647, 0.0556, 0.0555, 0.0533, 0.0661, 0.0622, 0.0644, 0.0642], device='cuda:1'), in_proj_covar=tensor([0.0097, 0.0079, 0.0096, 0.0096, 0.0107, 0.0073, 0.0102, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 09:58:09,139 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:13,856 INFO [train.py:873] (1/4) Epoch 3, batch 7300, loss[loss=0.216, simple_loss=0.2041, pruned_loss=0.114, over 7753.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2087, pruned_loss=0.1013, over 1883521.50 frames. ], batch size: 100, lr: 2.17e-02, grad_scale: 16.0 2022-12-07 09:58:31,585 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:49,753 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:59:06,176 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:59:22,139 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.55 vs. limit=5.0 2022-12-07 09:59:23,368 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:59:31,052 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.878e+02 3.644e+02 4.636e+02 1.010e+03, threshold=7.288e+02, percent-clipped=2.0 2022-12-07 09:59:34,518 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3769, 1.3330, 3.2630, 1.2873, 3.2758, 3.2831, 2.5080, 3.6085], device='cuda:1'), covar=tensor([0.0160, 0.2422, 0.0290, 0.2054, 0.0417, 0.0280, 0.0691, 0.0132], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0148, 0.0113, 0.0157, 0.0130, 0.0122, 0.0103, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 09:59:39,443 INFO [train.py:873] (1/4) Epoch 3, batch 7400, loss[loss=0.1922, simple_loss=0.2066, pruned_loss=0.08884, over 14439.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2104, pruned_loss=0.1024, over 1942614.94 frames. ], batch size: 53, lr: 2.17e-02, grad_scale: 16.0 2022-12-07 09:59:44,074 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6706, 0.7163, 0.7287, 0.7406, 0.5187, 0.3330, 0.3155, 0.6381], device='cuda:1'), covar=tensor([0.0113, 0.0059, 0.0099, 0.0082, 0.0219, 0.0267, 0.0075, 0.0277], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0010, 0.0011, 0.0010, 0.0011, 0.0013, 0.0012, 0.0016], device='cuda:1'), out_proj_covar=tensor([3.8831e-05, 3.6146e-05, 3.9371e-05, 3.6332e-05, 3.6751e-05, 4.1331e-05, 5.0286e-05, 5.3138e-05], device='cuda:1') 2022-12-07 09:59:45,705 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:05,236 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:19,735 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:25,315 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0767, 2.0862, 1.9740, 2.1936, 1.7495, 1.9602, 2.1064, 2.1701], device='cuda:1'), covar=tensor([0.0816, 0.0777, 0.0882, 0.0758, 0.1413, 0.0663, 0.0958, 0.0746], device='cuda:1'), in_proj_covar=tensor([0.0098, 0.0082, 0.0099, 0.0098, 0.0108, 0.0076, 0.0104, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:00:38,803 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:58,627 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.799e+02 3.655e+02 4.825e+02 8.691e+02, threshold=7.309e+02, percent-clipped=2.0 2022-12-07 10:01:06,422 INFO [train.py:873] (1/4) Epoch 3, batch 7500, loss[loss=0.2129, simple_loss=0.2185, pruned_loss=0.1036, over 14385.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2098, pruned_loss=0.1019, over 1948213.79 frames. ], batch size: 53, lr: 2.16e-02, grad_scale: 8.0 2022-12-07 10:01:07,382 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:01:18,014 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:01:44,222 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:02:33,691 INFO [train.py:873] (1/4) Epoch 4, batch 0, loss[loss=0.2411, simple_loss=0.252, pruned_loss=0.1151, over 14240.00 frames. ], tot_loss[loss=0.2411, simple_loss=0.252, pruned_loss=0.1151, over 14240.00 frames. ], batch size: 32, lr: 2.02e-02, grad_scale: 8.0 2022-12-07 10:02:33,692 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 10:02:40,698 INFO [train.py:905] (1/4) Epoch 4, validation: loss=0.1426, simple_loss=0.185, pruned_loss=0.0501, over 857387.00 frames. 2022-12-07 10:02:40,699 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 10:02:52,917 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:03:06,590 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.563e+01 2.095e+02 3.755e+02 5.061e+02 1.365e+03, threshold=7.510e+02, percent-clipped=11.0 2022-12-07 10:03:20,110 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3902, 4.1438, 4.0060, 4.4438, 4.0911, 3.5380, 4.4583, 4.3548], device='cuda:1'), covar=tensor([0.0651, 0.0536, 0.0575, 0.0644, 0.0576, 0.0544, 0.0572, 0.0890], device='cuda:1'), in_proj_covar=tensor([0.0097, 0.0081, 0.0098, 0.0098, 0.0107, 0.0074, 0.0103, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:03:28,540 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:08,184 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:08,914 INFO [train.py:873] (1/4) Epoch 4, batch 100, loss[loss=0.1858, simple_loss=0.1959, pruned_loss=0.08789, over 14095.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2079, pruned_loss=0.09572, over 890836.94 frames. ], batch size: 22, lr: 2.02e-02, grad_scale: 8.0 2022-12-07 10:04:14,172 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0420, 1.3179, 1.0274, 1.0042, 0.9669, 0.6789, 1.1215, 0.8478], device='cuda:1'), covar=tensor([0.0582, 0.0825, 0.0585, 0.0957, 0.0993, 0.0252, 0.0279, 0.1148], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0010, 0.0011, 0.0010, 0.0011, 0.0013, 0.0013, 0.0016], device='cuda:1'), out_proj_covar=tensor([3.8650e-05, 3.5692e-05, 4.0169e-05, 3.7306e-05, 3.8738e-05, 4.2223e-05, 5.1685e-05, 5.3537e-05], device='cuda:1') 2022-12-07 10:04:14,967 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:33,903 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.273e+02 2.903e+02 3.464e+02 4.472e+02 7.664e+02, threshold=6.927e+02, percent-clipped=2.0 2022-12-07 10:04:41,857 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:48,074 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:49,734 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:08,064 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:17,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2558, 1.3899, 2.5668, 1.3455, 2.4402, 2.4823, 1.8075, 2.5142], device='cuda:1'), covar=tensor([0.0199, 0.1600, 0.0171, 0.1523, 0.0239, 0.0246, 0.0760, 0.0172], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0150, 0.0113, 0.0159, 0.0131, 0.0121, 0.0102, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:05:22,205 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:29,630 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:35,015 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:35,727 INFO [train.py:873] (1/4) Epoch 4, batch 200, loss[loss=0.1963, simple_loss=0.2087, pruned_loss=0.09197, over 14538.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2044, pruned_loss=0.0948, over 1309432.22 frames. ], batch size: 43, lr: 2.01e-02, grad_scale: 8.0 2022-12-07 10:05:40,983 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:00,925 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.534e+02 3.313e+02 4.252e+02 7.669e+02, threshold=6.626e+02, percent-clipped=3.0 2022-12-07 10:06:03,426 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:22,003 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:33,414 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1088, 1.3347, 3.2236, 1.3145, 3.0005, 3.1327, 2.2434, 3.2494], device='cuda:1'), covar=tensor([0.0176, 0.2273, 0.0201, 0.2063, 0.0617, 0.0272, 0.0701, 0.0166], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0151, 0.0116, 0.0162, 0.0133, 0.0124, 0.0104, 0.0106], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:06:46,967 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 10:07:01,608 INFO [train.py:873] (1/4) Epoch 4, batch 300, loss[loss=0.2072, simple_loss=0.2049, pruned_loss=0.1048, over 14244.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2047, pruned_loss=0.09609, over 1546794.22 frames. ], batch size: 80, lr: 2.01e-02, grad_scale: 8.0 2022-12-07 10:07:08,466 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:07:26,196 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.872e+01 2.724e+02 3.752e+02 4.773e+02 9.337e+02, threshold=7.504e+02, percent-clipped=9.0 2022-12-07 10:07:48,254 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:07:55,259 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1254, 2.0881, 3.2645, 3.2821, 3.2901, 2.2663, 3.2093, 2.6117], device='cuda:1'), covar=tensor([0.0069, 0.0195, 0.0235, 0.0100, 0.0069, 0.0274, 0.0043, 0.0226], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0163, 0.0212, 0.0178, 0.0151, 0.0207, 0.0116, 0.0199], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:08:27,272 INFO [train.py:873] (1/4) Epoch 4, batch 400, loss[loss=0.1884, simple_loss=0.2091, pruned_loss=0.0839, over 14250.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2048, pruned_loss=0.09669, over 1707854.35 frames. ], batch size: 37, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:08:29,388 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:08:46,225 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:08:52,656 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.809e+02 3.789e+02 4.581e+02 1.009e+03, threshold=7.577e+02, percent-clipped=4.0 2022-12-07 10:09:22,101 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:28,091 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7937, 2.4812, 4.9058, 3.2441, 4.7860, 2.3951, 3.4150, 4.4318], device='cuda:1'), covar=tensor([0.0262, 0.5275, 0.0153, 1.0657, 0.0169, 0.3877, 0.1378, 0.0200], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0297, 0.0177, 0.0403, 0.0174, 0.0315, 0.0286, 0.0178], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:09:38,577 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:48,731 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:54,465 INFO [train.py:873] (1/4) Epoch 4, batch 500, loss[loss=0.2084, simple_loss=0.211, pruned_loss=0.1029, over 14381.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2061, pruned_loss=0.09714, over 1803733.00 frames. ], batch size: 55, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:10:02,960 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 10:10:05,611 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2022-12-07 10:10:06,711 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2022-12-07 10:10:19,854 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.033e+01 2.776e+02 3.736e+02 4.463e+02 9.726e+02, threshold=7.473e+02, percent-clipped=4.0 2022-12-07 10:10:20,855 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:10:31,882 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:13,480 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:18,881 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2022-12-07 10:11:20,903 INFO [train.py:873] (1/4) Epoch 4, batch 600, loss[loss=0.1604, simple_loss=0.1776, pruned_loss=0.07155, over 13964.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2055, pruned_loss=0.09712, over 1862067.37 frames. ], batch size: 20, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:11:23,691 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:27,810 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:45,413 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.720e+02 3.786e+02 4.640e+02 1.060e+03, threshold=7.571e+02, percent-clipped=5.0 2022-12-07 10:12:08,389 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:12:46,510 INFO [train.py:873] (1/4) Epoch 4, batch 700, loss[loss=0.2155, simple_loss=0.1963, pruned_loss=0.1173, over 3871.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2051, pruned_loss=0.09655, over 1893270.97 frames. ], batch size: 100, lr: 1.99e-02, grad_scale: 8.0 2022-12-07 10:13:07,681 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.37 vs. limit=2.0 2022-12-07 10:13:11,267 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.855e+02 3.544e+02 4.274e+02 7.481e+02, threshold=7.089e+02, percent-clipped=0.0 2022-12-07 10:13:30,235 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 10:13:40,610 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:13:47,794 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5405, 3.0035, 4.3026, 3.0275, 4.1327, 4.1296, 3.6585, 3.4549], device='cuda:1'), covar=tensor([0.0114, 0.1351, 0.0175, 0.0740, 0.0361, 0.0203, 0.0904, 0.0949], device='cuda:1'), in_proj_covar=tensor([0.0239, 0.0350, 0.0258, 0.0303, 0.0285, 0.0245, 0.0278, 0.0366], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:13:52,681 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:07,422 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:09,940 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4568, 4.9698, 5.0520, 5.4979, 5.2141, 4.3232, 5.6641, 5.4098], device='cuda:1'), covar=tensor([0.0674, 0.0541, 0.0519, 0.0545, 0.0627, 0.0394, 0.0517, 0.0683], device='cuda:1'), in_proj_covar=tensor([0.0104, 0.0085, 0.0105, 0.0103, 0.0112, 0.0077, 0.0110, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:14:12,390 INFO [train.py:873] (1/4) Epoch 4, batch 800, loss[loss=0.1908, simple_loss=0.2051, pruned_loss=0.0882, over 14263.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2056, pruned_loss=0.09715, over 1924254.47 frames. ], batch size: 63, lr: 1.99e-02, grad_scale: 8.0 2022-12-07 10:14:13,382 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:22,039 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:37,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 3.074e+02 3.598e+02 4.736e+02 1.088e+03, threshold=7.196e+02, percent-clipped=3.0 2022-12-07 10:14:48,566 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:05,959 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:15:26,575 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:36,454 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:38,303 INFO [train.py:873] (1/4) Epoch 4, batch 900, loss[loss=0.1846, simple_loss=0.1577, pruned_loss=0.1058, over 1249.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2059, pruned_loss=0.09679, over 1936483.96 frames. ], batch size: 100, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:16:03,091 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.303e+02 2.960e+02 3.909e+02 5.159e+02 1.247e+03, threshold=7.817e+02, percent-clipped=4.0 2022-12-07 10:17:03,224 INFO [train.py:873] (1/4) Epoch 4, batch 1000, loss[loss=0.2089, simple_loss=0.2105, pruned_loss=0.1037, over 14539.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2057, pruned_loss=0.09731, over 1954625.91 frames. ], batch size: 24, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:17:03,367 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8234, 4.5604, 4.9820, 4.1319, 4.8807, 5.2040, 1.8333, 4.6784], device='cuda:1'), covar=tensor([0.0141, 0.0227, 0.0343, 0.0380, 0.0204, 0.0080, 0.2531, 0.0183], device='cuda:1'), in_proj_covar=tensor([0.0113, 0.0121, 0.0114, 0.0095, 0.0154, 0.0107, 0.0145, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:17:18,362 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 10:17:28,303 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 2.746e+02 3.575e+02 4.376e+02 8.279e+02, threshold=7.149e+02, percent-clipped=2.0 2022-12-07 10:18:09,483 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:18:28,811 INFO [train.py:873] (1/4) Epoch 4, batch 1100, loss[loss=0.2211, simple_loss=0.2179, pruned_loss=0.1121, over 9464.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2055, pruned_loss=0.09677, over 2015237.23 frames. ], batch size: 100, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:18:35,283 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 10:18:50,542 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:18:53,789 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.927e+02 4.016e+02 5.307e+02 1.051e+03, threshold=8.031e+02, percent-clipped=8.0 2022-12-07 10:18:55,395 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2022-12-07 10:19:18,341 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:19:40,326 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.14 vs. limit=2.0 2022-12-07 10:19:43,544 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:19:53,953 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:19:55,485 INFO [train.py:873] (1/4) Epoch 4, batch 1200, loss[loss=0.2, simple_loss=0.2095, pruned_loss=0.09522, over 14473.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2063, pruned_loss=0.09767, over 2008211.95 frames. ], batch size: 24, lr: 1.97e-02, grad_scale: 8.0 2022-12-07 10:20:20,158 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 10:20:21,446 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.900e+02 3.609e+02 4.734e+02 8.045e+02, threshold=7.219e+02, percent-clipped=1.0 2022-12-07 10:20:24,896 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:20:25,910 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9647, 2.4092, 3.9646, 2.7808, 3.6810, 3.0447, 3.3030, 3.0728], device='cuda:1'), covar=tensor([0.0121, 0.1931, 0.0231, 0.0979, 0.0351, 0.0563, 0.1009, 0.1067], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0337, 0.0253, 0.0291, 0.0279, 0.0237, 0.0274, 0.0357], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:20:34,945 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:21:00,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2446, 3.1012, 2.7456, 2.8099, 3.1103, 3.0566, 3.1765, 3.1185], device='cuda:1'), covar=tensor([0.0751, 0.0879, 0.2126, 0.3029, 0.0785, 0.0794, 0.1195, 0.0980], device='cuda:1'), in_proj_covar=tensor([0.0234, 0.0200, 0.0291, 0.0380, 0.0230, 0.0265, 0.0275, 0.0228], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 10:21:20,673 INFO [train.py:873] (1/4) Epoch 4, batch 1300, loss[loss=0.2462, simple_loss=0.2261, pruned_loss=0.1332, over 8584.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2051, pruned_loss=0.09672, over 1994791.60 frames. ], batch size: 100, lr: 1.97e-02, grad_scale: 4.0 2022-12-07 10:21:46,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.737e+02 3.421e+02 4.208e+02 7.387e+02, threshold=6.842e+02, percent-clipped=1.0 2022-12-07 10:22:04,334 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:22:46,182 INFO [train.py:873] (1/4) Epoch 4, batch 1400, loss[loss=0.2018, simple_loss=0.211, pruned_loss=0.09628, over 14306.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.204, pruned_loss=0.09583, over 1952198.80 frames. ], batch size: 55, lr: 1.96e-02, grad_scale: 4.0 2022-12-07 10:22:55,434 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:23:12,193 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.769e+02 3.319e+02 4.343e+02 7.993e+02, threshold=6.638e+02, percent-clipped=1.0 2022-12-07 10:23:35,069 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:23:46,194 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 10:23:53,499 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1039, 1.7472, 2.2271, 2.3609, 1.9382, 1.6783, 2.3675, 2.0519], device='cuda:1'), covar=tensor([0.0047, 0.0093, 0.0071, 0.0037, 0.0048, 0.0165, 0.0044, 0.0072], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0169, 0.0228, 0.0189, 0.0154, 0.0213, 0.0125, 0.0206], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 10:24:03,724 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8427, 0.8660, 0.7590, 0.6604, 0.5953, 0.4547, 0.7842, 0.5424], device='cuda:1'), covar=tensor([0.0092, 0.0066, 0.0093, 0.0062, 0.0138, 0.0154, 0.0077, 0.0219], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0011, 0.0012, 0.0010, 0.0011, 0.0014, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([4.3472e-05, 4.0889e-05, 4.6471e-05, 3.8352e-05, 4.1418e-05, 4.9245e-05, 5.5594e-05, 6.0738e-05], device='cuda:1') 2022-12-07 10:24:11,899 INFO [train.py:873] (1/4) Epoch 4, batch 1500, loss[loss=0.2122, simple_loss=0.218, pruned_loss=0.1032, over 14663.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2045, pruned_loss=0.09687, over 1931041.83 frames. ], batch size: 33, lr: 1.96e-02, grad_scale: 4.0 2022-12-07 10:24:16,760 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:24:38,553 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.634e+02 3.340e+02 4.470e+02 8.988e+02, threshold=6.680e+02, percent-clipped=4.0 2022-12-07 10:25:07,509 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6800, 0.6262, 0.9953, 0.8230, 0.7299, 0.8393, 0.9883, 1.0956], device='cuda:1'), covar=tensor([0.0518, 0.1013, 0.0210, 0.0921, 0.0694, 0.0246, 0.0439, 0.0346], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0018, 0.0019, 0.0016, 0.0016, 0.0021, 0.0018, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.4530e-05, 5.9392e-05, 5.6362e-05, 5.5916e-05, 5.3832e-05, 6.5500e-05, 6.2841e-05, 5.4871e-05], device='cuda:1') 2022-12-07 10:25:34,451 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7688, 2.7663, 2.3977, 1.7267, 1.8346, 2.8698, 2.4645, 1.0490], device='cuda:1'), covar=tensor([0.3454, 0.1132, 0.2335, 0.1774, 0.1302, 0.0631, 0.1285, 0.3677], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0058, 0.0051, 0.0052, 0.0067, 0.0054, 0.0077, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:25:39,350 INFO [train.py:873] (1/4) Epoch 4, batch 1600, loss[loss=0.1942, simple_loss=0.1806, pruned_loss=0.1039, over 2615.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2042, pruned_loss=0.09582, over 1976708.89 frames. ], batch size: 100, lr: 1.96e-02, grad_scale: 8.0 2022-12-07 10:26:05,382 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.687e+02 3.404e+02 4.412e+02 1.604e+03, threshold=6.807e+02, percent-clipped=2.0 2022-12-07 10:26:33,599 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0099, 2.8300, 2.3477, 1.5801, 2.5540, 2.6803, 2.9674, 2.0796], device='cuda:1'), covar=tensor([0.0573, 0.3647, 0.1738, 0.4014, 0.1117, 0.0530, 0.1023, 0.2087], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0231, 0.0107, 0.0132, 0.0099, 0.0093, 0.0080, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:27:05,336 INFO [train.py:873] (1/4) Epoch 4, batch 1700, loss[loss=0.1905, simple_loss=0.1764, pruned_loss=0.1023, over 3842.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.204, pruned_loss=0.09566, over 1955097.33 frames. ], batch size: 100, lr: 1.95e-02, grad_scale: 8.0 2022-12-07 10:27:10,460 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:27:31,010 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.793e+02 3.554e+02 4.451e+02 1.467e+03, threshold=7.108e+02, percent-clipped=5.0 2022-12-07 10:27:49,552 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5560, 5.3649, 5.1417, 5.8255, 5.3254, 5.0054, 5.7856, 5.6993], device='cuda:1'), covar=tensor([0.0623, 0.0365, 0.0540, 0.0388, 0.0492, 0.0199, 0.0476, 0.0450], device='cuda:1'), in_proj_covar=tensor([0.0104, 0.0083, 0.0104, 0.0100, 0.0111, 0.0076, 0.0110, 0.0101], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:28:30,189 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0688, 3.8267, 3.4870, 3.5855, 3.7711, 3.8872, 4.0570, 3.9734], device='cuda:1'), covar=tensor([0.0678, 0.0657, 0.1495, 0.2484, 0.0734, 0.0633, 0.0796, 0.0804], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0210, 0.0305, 0.0382, 0.0237, 0.0275, 0.0277, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 10:28:31,821 INFO [train.py:873] (1/4) Epoch 4, batch 1800, loss[loss=0.1793, simple_loss=0.1516, pruned_loss=0.1035, over 2631.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.205, pruned_loss=0.09617, over 1996963.70 frames. ], batch size: 100, lr: 1.95e-02, grad_scale: 8.0 2022-12-07 10:28:42,738 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 10:28:44,304 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2022-12-07 10:28:55,813 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 10:28:57,764 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.645e+02 3.516e+02 4.741e+02 1.240e+03, threshold=7.031e+02, percent-clipped=7.0 2022-12-07 10:29:04,970 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0277, 2.1916, 1.8189, 1.8492, 1.3915, 1.9954, 1.8460, 0.7274], device='cuda:1'), covar=tensor([0.3506, 0.0979, 0.2423, 0.1288, 0.1908, 0.0638, 0.1348, 0.4819], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0060, 0.0053, 0.0054, 0.0071, 0.0056, 0.0079, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:29:08,910 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1619, 4.8919, 4.7887, 5.2066, 4.8230, 4.4922, 5.2249, 5.1426], device='cuda:1'), covar=tensor([0.0565, 0.0463, 0.0540, 0.0525, 0.0529, 0.0351, 0.0514, 0.0571], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0084, 0.0105, 0.0103, 0.0112, 0.0078, 0.0112, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:29:23,749 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:29:25,438 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8178, 2.7777, 1.9391, 2.9228, 2.6354, 2.8910, 2.4959, 2.1946], device='cuda:1'), covar=tensor([0.0388, 0.1072, 0.3526, 0.0263, 0.0557, 0.0541, 0.1036, 0.3542], device='cuda:1'), in_proj_covar=tensor([0.0216, 0.0291, 0.0322, 0.0187, 0.0235, 0.0223, 0.0259, 0.0319], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:29:43,023 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:29:58,461 INFO [train.py:873] (1/4) Epoch 4, batch 1900, loss[loss=0.1783, simple_loss=0.1936, pruned_loss=0.08154, over 14047.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2042, pruned_loss=0.09577, over 1948065.10 frames. ], batch size: 19, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:30:17,339 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:30:21,519 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3206, 2.3320, 1.7795, 2.4931, 2.2821, 2.4002, 2.0878, 1.9135], device='cuda:1'), covar=tensor([0.0199, 0.0676, 0.1762, 0.0198, 0.0510, 0.0226, 0.0908, 0.1428], device='cuda:1'), in_proj_covar=tensor([0.0214, 0.0294, 0.0324, 0.0187, 0.0234, 0.0225, 0.0264, 0.0323], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:30:24,686 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.734e+02 3.498e+02 4.101e+02 8.612e+02, threshold=6.995e+02, percent-clipped=1.0 2022-12-07 10:30:36,043 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:31:09,580 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.29 vs. limit=2.0 2022-12-07 10:31:24,628 INFO [train.py:873] (1/4) Epoch 4, batch 2000, loss[loss=0.1921, simple_loss=0.1691, pruned_loss=0.1076, over 2727.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2051, pruned_loss=0.09636, over 1954082.58 frames. ], batch size: 100, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:31:29,932 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:31:39,450 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8500, 2.6948, 2.9131, 2.6878, 2.6871, 2.6700, 1.2471, 2.5676], device='cuda:1'), covar=tensor([0.0203, 0.0277, 0.0308, 0.0281, 0.0283, 0.0454, 0.2361, 0.0259], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0124, 0.0115, 0.0100, 0.0155, 0.0107, 0.0147, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:31:50,653 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.930e+02 3.788e+02 4.800e+02 1.678e+03, threshold=7.576e+02, percent-clipped=5.0 2022-12-07 10:32:08,579 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0333, 1.9616, 2.0162, 2.0119, 1.9368, 1.8100, 1.0297, 1.7684], device='cuda:1'), covar=tensor([0.0305, 0.0258, 0.0424, 0.0212, 0.0310, 0.0718, 0.1817, 0.0352], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0124, 0.0115, 0.0099, 0.0155, 0.0109, 0.0147, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:32:11,156 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:32:27,584 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7733, 2.3556, 4.7555, 3.0970, 4.4262, 2.4610, 3.5242, 4.4216], device='cuda:1'), covar=tensor([0.0319, 0.5659, 0.0436, 0.9716, 0.0366, 0.4420, 0.1164, 0.0145], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0296, 0.0177, 0.0396, 0.0176, 0.0306, 0.0279, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0004, 0.0002], device='cuda:1') 2022-12-07 10:32:32,510 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:32:51,153 INFO [train.py:873] (1/4) Epoch 4, batch 2100, loss[loss=0.1977, simple_loss=0.2034, pruned_loss=0.09598, over 10349.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2041, pruned_loss=0.09588, over 1933445.18 frames. ], batch size: 100, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:33:17,441 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.608e+02 3.300e+02 4.069e+02 1.028e+03, threshold=6.599e+02, percent-clipped=1.0 2022-12-07 10:33:25,438 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:33:44,808 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:18,570 INFO [train.py:873] (1/4) Epoch 4, batch 2200, loss[loss=0.2262, simple_loss=0.1967, pruned_loss=0.1278, over 3898.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2047, pruned_loss=0.09729, over 1901575.56 frames. ], batch size: 100, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:34:31,890 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:37,055 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:37,875 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0566, 0.8223, 1.2452, 0.8994, 1.0775, 0.5768, 0.9557, 1.1942], device='cuda:1'), covar=tensor([0.0455, 0.0453, 0.0145, 0.0818, 0.0234, 0.0223, 0.0537, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0018, 0.0019, 0.0016, 0.0016, 0.0022, 0.0019, 0.0018], device='cuda:1'), out_proj_covar=tensor([5.4897e-05, 5.9647e-05, 5.8243e-05, 5.6469e-05, 5.3853e-05, 6.7879e-05, 6.3948e-05, 5.7896e-05], device='cuda:1') 2022-12-07 10:34:43,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.748e+02 3.566e+02 4.781e+02 9.343e+02, threshold=7.131e+02, percent-clipped=9.0 2022-12-07 10:34:50,870 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:35:00,085 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2022-12-07 10:35:09,903 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9639, 2.4057, 4.8925, 3.3553, 4.5799, 2.2620, 3.5966, 4.4865], device='cuda:1'), covar=tensor([0.0215, 0.4969, 0.0204, 0.8653, 0.0211, 0.4060, 0.1134, 0.0193], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0288, 0.0173, 0.0385, 0.0174, 0.0301, 0.0272, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:35:44,687 INFO [train.py:873] (1/4) Epoch 4, batch 2300, loss[loss=0.1786, simple_loss=0.1989, pruned_loss=0.07912, over 14287.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2042, pruned_loss=0.09717, over 1878469.78 frames. ], batch size: 44, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:35:51,923 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:36:14,179 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.734e+02 3.348e+02 4.422e+02 1.416e+03, threshold=6.696e+02, percent-clipped=4.0 2022-12-07 10:36:48,419 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:37:01,968 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7133, 0.6908, 0.3490, 0.8297, 0.6109, 0.2727, 0.6632, 0.7514], device='cuda:1'), covar=tensor([0.0081, 0.0045, 0.0037, 0.0070, 0.0049, 0.0046, 0.0219, 0.0076], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0019, 0.0020, 0.0016, 0.0016, 0.0022, 0.0019, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.6363e-05, 6.0946e-05, 5.9012e-05, 5.7472e-05, 5.5022e-05, 6.8904e-05, 6.5995e-05, 5.7544e-05], device='cuda:1') 2022-12-07 10:37:15,454 INFO [train.py:873] (1/4) Epoch 4, batch 2400, loss[loss=0.2148, simple_loss=0.2135, pruned_loss=0.108, over 12003.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2047, pruned_loss=0.09673, over 1937042.96 frames. ], batch size: 100, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:37:41,018 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.617e+02 3.623e+02 4.580e+02 1.018e+03, threshold=7.246e+02, percent-clipped=1.0 2022-12-07 10:37:44,899 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:37:49,154 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6328, 3.9161, 3.2267, 5.0370, 4.3591, 4.7922, 3.7693, 3.4685], device='cuda:1'), covar=tensor([0.0502, 0.1024, 0.3516, 0.0280, 0.0426, 0.0804, 0.0899, 0.2839], device='cuda:1'), in_proj_covar=tensor([0.0220, 0.0290, 0.0327, 0.0189, 0.0234, 0.0226, 0.0261, 0.0315], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:38:41,251 INFO [train.py:873] (1/4) Epoch 4, batch 2500, loss[loss=0.1822, simple_loss=0.1993, pruned_loss=0.08252, over 14244.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2041, pruned_loss=0.09602, over 1921271.78 frames. ], batch size: 69, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:38:55,771 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:38:56,530 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:07,387 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.934e+02 3.822e+02 4.724e+02 8.736e+02, threshold=7.644e+02, percent-clipped=3.0 2022-12-07 10:39:14,166 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:16,666 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8480, 4.8296, 5.7848, 5.7384, 5.5031, 4.6984, 5.8770, 4.8472], device='cuda:1'), covar=tensor([0.0738, 0.2739, 0.0524, 0.0678, 0.0872, 0.0586, 0.0533, 0.1078], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0196, 0.0128, 0.0122, 0.0130, 0.0108, 0.0191, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:39:36,299 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1376, 1.3811, 1.4455, 1.3234, 1.1124, 1.2669, 1.1915, 0.8343], device='cuda:1'), covar=tensor([0.2630, 0.1144, 0.0987, 0.0575, 0.0982, 0.0757, 0.1820, 0.3227], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0061, 0.0050, 0.0052, 0.0069, 0.0056, 0.0079, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:39:37,051 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:56,075 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:40:08,587 INFO [train.py:873] (1/4) Epoch 4, batch 2600, loss[loss=0.1976, simple_loss=0.2057, pruned_loss=0.09472, over 12752.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2042, pruned_loss=0.0965, over 1895747.14 frames. ], batch size: 100, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:40:11,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.78 vs. limit=5.0 2022-12-07 10:40:33,522 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2022-12-07 10:40:34,421 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.970e+01 2.618e+02 3.461e+02 4.480e+02 1.117e+03, threshold=6.922e+02, percent-clipped=3.0 2022-12-07 10:41:03,029 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:41:23,580 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 10:41:33,623 INFO [train.py:873] (1/4) Epoch 4, batch 2700, loss[loss=0.1745, simple_loss=0.194, pruned_loss=0.07749, over 13971.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2042, pruned_loss=0.09606, over 1967466.25 frames. ], batch size: 22, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:41:59,634 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.867e+02 3.536e+02 4.473e+02 9.787e+02, threshold=7.071e+02, percent-clipped=5.0 2022-12-07 10:42:03,215 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:42:05,756 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7169, 1.6005, 1.9652, 1.5984, 1.4170, 1.9437, 1.2346, 1.8290], device='cuda:1'), covar=tensor([0.0837, 0.1805, 0.0315, 0.1109, 0.1484, 0.0418, 0.2300, 0.0386], device='cuda:1'), in_proj_covar=tensor([0.0066, 0.0072, 0.0065, 0.0077, 0.0086, 0.0062, 0.0134, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:42:09,240 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:42:30,688 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=11.89 vs. limit=5.0 2022-12-07 10:42:44,237 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:43:00,508 INFO [train.py:873] (1/4) Epoch 4, batch 2800, loss[loss=0.1928, simple_loss=0.2019, pruned_loss=0.09181, over 14420.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2044, pruned_loss=0.09551, over 1996380.53 frames. ], batch size: 73, lr: 1.91e-02, grad_scale: 8.0 2022-12-07 10:43:02,444 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:43:14,953 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:43:25,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.496e+02 3.123e+02 4.474e+02 9.387e+02, threshold=6.246e+02, percent-clipped=3.0 2022-12-07 10:43:37,643 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7057, 2.1112, 5.1882, 4.7994, 4.7528, 5.1209, 5.0776, 5.2385], device='cuda:1'), covar=tensor([0.0925, 0.1194, 0.0045, 0.0099, 0.0096, 0.0070, 0.0041, 0.0073], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0152, 0.0089, 0.0123, 0.0104, 0.0109, 0.0080, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:43:46,024 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8862, 1.4091, 2.6962, 2.5720, 2.7414, 2.6394, 1.9107, 2.7313], device='cuda:1'), covar=tensor([0.0796, 0.0987, 0.0079, 0.0198, 0.0143, 0.0094, 0.0265, 0.0114], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0153, 0.0090, 0.0124, 0.0105, 0.0110, 0.0081, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:43:53,501 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9575, 0.8344, 0.6214, 1.0201, 0.8696, 0.4101, 0.7824, 0.9526], device='cuda:1'), covar=tensor([0.0608, 0.0372, 0.0129, 0.0197, 0.0159, 0.0182, 0.0343, 0.0249], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0019, 0.0020, 0.0016, 0.0017, 0.0022, 0.0018, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.6890e-05, 6.2237e-05, 5.9443e-05, 5.8136e-05, 5.5359e-05, 7.0917e-05, 6.5677e-05, 5.6540e-05], device='cuda:1') 2022-12-07 10:43:55,790 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:44:26,291 INFO [train.py:873] (1/4) Epoch 4, batch 2900, loss[loss=0.1597, simple_loss=0.1506, pruned_loss=0.08441, over 2593.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2034, pruned_loss=0.09519, over 1948145.61 frames. ], batch size: 100, lr: 1.91e-02, grad_scale: 8.0 2022-12-07 10:44:34,213 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 10:44:52,742 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 2.734e+02 3.612e+02 4.564e+02 7.787e+02, threshold=7.225e+02, percent-clipped=1.0 2022-12-07 10:45:22,123 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:45:33,517 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1692, 3.1257, 4.0043, 2.7277, 2.7309, 2.6006, 1.5546, 3.1324], device='cuda:1'), covar=tensor([0.0872, 0.0850, 0.0429, 0.1829, 0.1534, 0.1378, 0.4846, 0.0919], device='cuda:1'), in_proj_covar=tensor([0.0066, 0.0071, 0.0066, 0.0079, 0.0089, 0.0063, 0.0136, 0.0071], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:45:47,572 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3177, 3.0387, 3.4052, 3.0221, 3.2661, 2.9020, 1.2316, 3.1489], device='cuda:1'), covar=tensor([0.0218, 0.0369, 0.0419, 0.0401, 0.0304, 0.0543, 0.2928, 0.0263], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0128, 0.0114, 0.0099, 0.0155, 0.0109, 0.0146, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:45:53,684 INFO [train.py:873] (1/4) Epoch 4, batch 3000, loss[loss=0.1868, simple_loss=0.1894, pruned_loss=0.09206, over 5944.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2023, pruned_loss=0.09417, over 1914933.62 frames. ], batch size: 100, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:45:53,685 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 10:46:03,274 INFO [train.py:905] (1/4) Epoch 4, validation: loss=0.1268, simple_loss=0.1698, pruned_loss=0.0419, over 857387.00 frames. 2022-12-07 10:46:03,275 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 10:46:14,151 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:46:29,651 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.570e+02 3.689e+02 4.619e+02 9.776e+02, threshold=7.378e+02, percent-clipped=3.0 2022-12-07 10:47:28,446 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:47:28,976 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 10:47:30,950 INFO [train.py:873] (1/4) Epoch 4, batch 3100, loss[loss=0.166, simple_loss=0.1901, pruned_loss=0.07098, over 14507.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2027, pruned_loss=0.09433, over 1930575.80 frames. ], batch size: 49, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:47:57,004 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.578e+02 3.545e+02 4.492e+02 7.913e+02, threshold=7.089e+02, percent-clipped=1.0 2022-12-07 10:48:33,325 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:48:57,814 INFO [train.py:873] (1/4) Epoch 4, batch 3200, loss[loss=0.2303, simple_loss=0.1934, pruned_loss=0.1336, over 1295.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2034, pruned_loss=0.09505, over 1943161.12 frames. ], batch size: 100, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:49:24,095 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.801e+02 3.510e+02 4.555e+02 1.491e+03, threshold=7.021e+02, percent-clipped=7.0 2022-12-07 10:49:25,926 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:49:49,639 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1768, 1.3917, 3.1738, 1.2971, 3.0664, 3.2423, 2.1945, 3.3536], device='cuda:1'), covar=tensor([0.0194, 0.2244, 0.0316, 0.1990, 0.0603, 0.0252, 0.0631, 0.0163], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0152, 0.0122, 0.0162, 0.0136, 0.0130, 0.0112, 0.0108], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:49:54,638 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6829, 1.3420, 1.8435, 2.0966, 1.3699, 1.4848, 1.9187, 1.7947], device='cuda:1'), covar=tensor([0.0037, 0.0087, 0.0031, 0.0022, 0.0058, 0.0098, 0.0034, 0.0034], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0173, 0.0239, 0.0197, 0.0161, 0.0221, 0.0135, 0.0215], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 10:50:25,175 INFO [train.py:873] (1/4) Epoch 4, batch 3300, loss[loss=0.2346, simple_loss=0.2247, pruned_loss=0.1223, over 14127.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2038, pruned_loss=0.09563, over 1932921.25 frames. ], batch size: 99, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:50:44,330 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8622, 1.8844, 2.8925, 2.8077, 2.7729, 2.0099, 2.6841, 2.1959], device='cuda:1'), covar=tensor([0.0070, 0.0205, 0.0156, 0.0098, 0.0071, 0.0329, 0.0047, 0.0254], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0172, 0.0237, 0.0195, 0.0159, 0.0220, 0.0133, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 10:50:51,957 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.499e+02 3.235e+02 4.436e+02 1.031e+03, threshold=6.470e+02, percent-clipped=3.0 2022-12-07 10:51:48,820 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:51:51,018 INFO [train.py:873] (1/4) Epoch 4, batch 3400, loss[loss=0.197, simple_loss=0.1987, pruned_loss=0.09765, over 5994.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.202, pruned_loss=0.09386, over 1955255.27 frames. ], batch size: 100, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:51:58,607 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 10:52:04,567 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:52:18,305 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.979e+02 3.585e+02 4.923e+02 1.213e+03, threshold=7.170e+02, percent-clipped=7.0 2022-12-07 10:52:30,047 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:52:31,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.5630, 1.2125, 1.3056, 1.1783, 1.0605, 1.2563, 1.1125, 0.7128], device='cuda:1'), covar=tensor([0.2694, 0.0744, 0.0426, 0.0286, 0.0659, 0.0251, 0.1257, 0.1084], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0060, 0.0050, 0.0053, 0.0069, 0.0053, 0.0078, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:52:51,106 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 10:52:57,404 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:53:01,647 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:53:18,122 INFO [train.py:873] (1/4) Epoch 4, batch 3500, loss[loss=0.2398, simple_loss=0.2268, pruned_loss=0.1264, over 8609.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2019, pruned_loss=0.09424, over 1954056.11 frames. ], batch size: 100, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:53:41,613 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 10:53:44,755 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.647e+02 3.463e+02 4.115e+02 9.483e+02, threshold=6.926e+02, percent-clipped=4.0 2022-12-07 10:53:54,497 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:54:11,998 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:54:44,198 INFO [train.py:873] (1/4) Epoch 4, batch 3600, loss[loss=0.1687, simple_loss=0.1517, pruned_loss=0.09283, over 2655.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2017, pruned_loss=0.09412, over 1904050.79 frames. ], batch size: 100, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:54:59,238 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7081, 1.2707, 2.3864, 2.3372, 2.5012, 2.4077, 1.8534, 2.4690], device='cuda:1'), covar=tensor([0.0562, 0.0927, 0.0099, 0.0187, 0.0151, 0.0086, 0.0293, 0.0118], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0153, 0.0093, 0.0126, 0.0107, 0.0111, 0.0083, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:55:05,371 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:55:10,969 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.617e+02 3.267e+02 4.043e+02 9.670e+02, threshold=6.534e+02, percent-clipped=3.0 2022-12-07 10:55:24,794 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0820, 2.7977, 2.5398, 1.8910, 2.5805, 2.8784, 3.2659, 2.2680], device='cuda:1'), covar=tensor([0.0487, 0.2855, 0.1165, 0.2640, 0.1197, 0.0631, 0.0541, 0.1739], device='cuda:1'), in_proj_covar=tensor([0.0092, 0.0219, 0.0107, 0.0128, 0.0099, 0.0092, 0.0080, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:55:50,241 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3189, 2.5679, 3.9266, 2.7254, 4.0255, 3.7653, 3.6786, 3.1668], device='cuda:1'), covar=tensor([0.0186, 0.2146, 0.0382, 0.1228, 0.0336, 0.0373, 0.1181, 0.1641], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0342, 0.0284, 0.0307, 0.0295, 0.0253, 0.0293, 0.0359], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 10:56:10,704 INFO [train.py:873] (1/4) Epoch 4, batch 3700, loss[loss=0.1993, simple_loss=0.207, pruned_loss=0.0958, over 14213.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2024, pruned_loss=0.09424, over 1963219.58 frames. ], batch size: 99, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:56:33,792 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:56:37,277 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.639e+02 3.495e+02 4.493e+02 9.568e+02, threshold=6.990e+02, percent-clipped=8.0 2022-12-07 10:57:11,874 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:57:25,470 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.58 vs. limit=5.0 2022-12-07 10:57:26,698 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:57:30,037 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2953, 1.8317, 4.2577, 2.9442, 3.8886, 1.8213, 2.9638, 3.9138], device='cuda:1'), covar=tensor([0.0296, 0.5415, 0.0198, 0.7465, 0.0327, 0.4205, 0.1405, 0.0222], device='cuda:1'), in_proj_covar=tensor([0.0222, 0.0278, 0.0170, 0.0367, 0.0173, 0.0283, 0.0259, 0.0167], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:57:36,083 INFO [train.py:873] (1/4) Epoch 4, batch 3800, loss[loss=0.1875, simple_loss=0.1965, pruned_loss=0.08923, over 13544.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2021, pruned_loss=0.09381, over 1929393.28 frames. ], batch size: 100, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:58:00,301 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:58:03,348 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.752e+02 3.266e+02 4.284e+02 8.876e+02, threshold=6.533e+02, percent-clipped=2.0 2022-12-07 10:58:08,562 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:58:08,970 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.41 vs. limit=5.0 2022-12-07 10:58:12,491 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 10:58:36,570 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5471, 1.6315, 3.5826, 2.3734, 3.5976, 1.6916, 2.1543, 3.4335], device='cuda:1'), covar=tensor([0.0643, 0.6475, 0.0338, 1.1215, 0.0372, 0.4765, 0.2420, 0.0463], device='cuda:1'), in_proj_covar=tensor([0.0226, 0.0279, 0.0174, 0.0377, 0.0175, 0.0292, 0.0263, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:1') 2022-12-07 10:58:37,301 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7684, 3.5777, 3.4646, 3.9661, 3.4208, 3.1667, 3.8530, 3.8547], device='cuda:1'), covar=tensor([0.0734, 0.0670, 0.0744, 0.0521, 0.0863, 0.0637, 0.0714, 0.0635], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0087, 0.0105, 0.0104, 0.0114, 0.0081, 0.0115, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:1') 2022-12-07 10:58:41,458 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:59:03,303 INFO [train.py:873] (1/4) Epoch 4, batch 3900, loss[loss=0.2273, simple_loss=0.2125, pruned_loss=0.121, over 7795.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2011, pruned_loss=0.09286, over 2006108.80 frames. ], batch size: 100, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 10:59:04,224 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0294, 2.7991, 2.6009, 2.7118, 2.9095, 2.8138, 2.9713, 2.9486], device='cuda:1'), covar=tensor([0.0728, 0.0786, 0.1701, 0.2116, 0.0737, 0.0832, 0.1125, 0.0787], device='cuda:1'), in_proj_covar=tensor([0.0248, 0.0215, 0.0307, 0.0388, 0.0238, 0.0279, 0.0289, 0.0226], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 10:59:14,365 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1941, 3.6589, 4.1283, 3.5446, 4.0164, 4.2756, 1.5930, 3.7890], device='cuda:1'), covar=tensor([0.0250, 0.0465, 0.0653, 0.0638, 0.0401, 0.0272, 0.3250, 0.0346], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0126, 0.0119, 0.0099, 0.0156, 0.0111, 0.0148, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 10:59:19,350 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:59:29,337 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.794e+02 3.682e+02 4.681e+02 9.431e+02, threshold=7.364e+02, percent-clipped=3.0 2022-12-07 10:59:36,519 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 10:59:43,090 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6206, 2.0116, 2.6653, 2.2084, 2.7439, 2.4891, 2.4154, 2.2698], device='cuda:1'), covar=tensor([0.0155, 0.1282, 0.0261, 0.0739, 0.0218, 0.0319, 0.0227, 0.0921], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0353, 0.0288, 0.0316, 0.0304, 0.0260, 0.0302, 0.0371], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:00:12,362 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 11:00:16,812 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 11:00:28,849 INFO [train.py:873] (1/4) Epoch 4, batch 4000, loss[loss=0.1977, simple_loss=0.1905, pruned_loss=0.1024, over 4981.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2013, pruned_loss=0.09304, over 1990408.39 frames. ], batch size: 100, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 11:00:40,043 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8441, 1.7957, 2.1999, 1.6674, 1.4822, 1.9802, 1.3321, 1.9177], device='cuda:1'), covar=tensor([0.1140, 0.1298, 0.0578, 0.1902, 0.2785, 0.0910, 0.4584, 0.1126], device='cuda:1'), in_proj_covar=tensor([0.0066, 0.0072, 0.0065, 0.0077, 0.0093, 0.0060, 0.0133, 0.0069], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:00:47,683 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 11:00:55,534 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.732e+02 2.728e+02 3.637e+02 4.905e+02 8.607e+02, threshold=7.274e+02, percent-clipped=2.0 2022-12-07 11:01:15,014 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0851, 1.8601, 4.4240, 4.2660, 4.2358, 4.5649, 4.0969, 4.5936], device='cuda:1'), covar=tensor([0.1040, 0.1116, 0.0071, 0.0095, 0.0096, 0.0077, 0.0117, 0.0061], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0153, 0.0093, 0.0125, 0.0107, 0.0112, 0.0080, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:01:21,148 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1579, 1.9033, 3.2904, 2.2392, 3.1030, 1.8091, 2.4919, 2.8735], device='cuda:1'), covar=tensor([0.0670, 0.4824, 0.0373, 0.8592, 0.0322, 0.4095, 0.1692, 0.0345], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0284, 0.0178, 0.0387, 0.0177, 0.0292, 0.0264, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 11:01:28,451 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6793, 2.8390, 4.3153, 3.1890, 4.2684, 4.3085, 3.9773, 3.7069], device='cuda:1'), covar=tensor([0.0158, 0.2049, 0.0314, 0.1160, 0.0388, 0.0307, 0.1288, 0.1231], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0353, 0.0289, 0.0313, 0.0304, 0.0258, 0.0302, 0.0367], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:01:29,166 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:01:39,508 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:01:53,836 INFO [train.py:873] (1/4) Epoch 4, batch 4100, loss[loss=0.2322, simple_loss=0.2293, pruned_loss=0.1176, over 10317.00 frames. ], tot_loss[loss=0.1931, simple_loss=0.2011, pruned_loss=0.09251, over 1997374.83 frames. ], batch size: 100, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 11:02:10,420 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:02:20,838 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.498e+02 3.224e+02 4.280e+02 1.572e+03, threshold=6.447e+02, percent-clipped=4.0 2022-12-07 11:02:26,108 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:02:33,633 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6830, 1.6482, 2.0794, 1.7180, 2.0591, 1.8218, 1.6757, 1.9230], device='cuda:1'), covar=tensor([0.0191, 0.0504, 0.0080, 0.0113, 0.0069, 0.0110, 0.0065, 0.0125], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0344, 0.0283, 0.0305, 0.0302, 0.0253, 0.0297, 0.0362], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:02:39,052 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 11:02:55,334 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2022-12-07 11:03:07,342 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:03:19,883 INFO [train.py:873] (1/4) Epoch 4, batch 4200, loss[loss=0.2772, simple_loss=0.2268, pruned_loss=0.1637, over 1211.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2009, pruned_loss=0.09251, over 1942510.73 frames. ], batch size: 100, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:03:34,683 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:03:36,743 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:03:46,350 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.657e+02 3.822e+02 4.895e+02 8.860e+02, threshold=7.643e+02, percent-clipped=9.0 2022-12-07 11:04:04,019 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6713, 2.8792, 4.4799, 3.1982, 4.2962, 4.5016, 3.8479, 3.6378], device='cuda:1'), covar=tensor([0.0137, 0.1596, 0.0224, 0.0939, 0.0320, 0.0173, 0.1671, 0.0933], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0348, 0.0288, 0.0310, 0.0305, 0.0257, 0.0303, 0.0366], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:04:11,953 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4689, 2.1695, 4.9016, 4.4655, 4.4339, 4.8562, 4.6592, 4.9061], device='cuda:1'), covar=tensor([0.1132, 0.1268, 0.0096, 0.0111, 0.0150, 0.0120, 0.0077, 0.0119], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0151, 0.0092, 0.0124, 0.0106, 0.0111, 0.0080, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:04:13,524 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:04:16,900 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:04:26,246 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:04:45,368 INFO [train.py:873] (1/4) Epoch 4, batch 4300, loss[loss=0.2183, simple_loss=0.2076, pruned_loss=0.1144, over 7789.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2026, pruned_loss=0.09449, over 1931324.51 frames. ], batch size: 100, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:05:05,658 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:05:10,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6765, 4.4581, 4.0934, 4.2135, 4.2784, 4.4533, 4.6030, 4.5881], device='cuda:1'), covar=tensor([0.0610, 0.0441, 0.1701, 0.2240, 0.0656, 0.0559, 0.0719, 0.0686], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0213, 0.0311, 0.0394, 0.0241, 0.0280, 0.0293, 0.0231], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:05:11,530 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.870e+02 3.578e+02 4.388e+02 9.558e+02, threshold=7.157e+02, percent-clipped=3.0 2022-12-07 11:05:57,152 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:05:57,764 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 11:06:11,314 INFO [train.py:873] (1/4) Epoch 4, batch 4400, loss[loss=0.1842, simple_loss=0.2022, pruned_loss=0.08308, over 14265.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2024, pruned_loss=0.09432, over 1909581.00 frames. ], batch size: 31, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:06:14,572 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2022-12-07 11:06:32,708 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0325, 1.9927, 1.5818, 2.1000, 1.8205, 2.0467, 1.8233, 1.7429], device='cuda:1'), covar=tensor([0.0297, 0.0466, 0.1114, 0.0092, 0.0383, 0.0173, 0.0512, 0.0320], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0292, 0.0315, 0.0187, 0.0239, 0.0230, 0.0258, 0.0309], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:1') 2022-12-07 11:06:38,400 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.559e+02 3.461e+02 4.606e+02 7.213e+02, threshold=6.922e+02, percent-clipped=1.0 2022-12-07 11:06:38,480 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:07:04,497 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:07:37,302 INFO [train.py:873] (1/4) Epoch 4, batch 4500, loss[loss=0.1583, simple_loss=0.1822, pruned_loss=0.0672, over 14279.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2004, pruned_loss=0.09187, over 1929268.91 frames. ], batch size: 44, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:07:57,256 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:08:04,791 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.722e+01 2.647e+02 3.341e+02 4.493e+02 1.287e+03, threshold=6.683e+02, percent-clipped=3.0 2022-12-07 11:08:35,856 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1518, 2.7996, 2.9794, 2.2016, 1.8590, 2.8752, 2.5718, 1.6465], device='cuda:1'), covar=tensor([0.3286, 0.0748, 0.1099, 0.1478, 0.1253, 0.1123, 0.1753, 0.3281], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0058, 0.0050, 0.0051, 0.0068, 0.0052, 0.0080, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:08:40,900 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 11:09:00,150 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8476, 2.8265, 3.4135, 2.5717, 2.3840, 2.8374, 1.4521, 2.8838], device='cuda:1'), covar=tensor([0.1127, 0.0700, 0.0690, 0.1706, 0.1794, 0.1056, 0.5092, 0.1136], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0075, 0.0070, 0.0083, 0.0097, 0.0066, 0.0141, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:09:04,082 INFO [train.py:873] (1/4) Epoch 4, batch 4600, loss[loss=0.2083, simple_loss=0.2109, pruned_loss=0.1029, over 7803.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2007, pruned_loss=0.09164, over 1957737.64 frames. ], batch size: 100, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:09:20,180 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:09:25,323 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 11:09:30,612 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 3.048e+02 4.012e+02 4.961e+02 8.817e+02, threshold=8.024e+02, percent-clipped=5.0 2022-12-07 11:10:23,531 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6453, 4.2900, 4.2275, 4.7149, 4.3474, 3.7823, 4.7722, 4.6183], device='cuda:1'), covar=tensor([0.0628, 0.0492, 0.0535, 0.0476, 0.0635, 0.0552, 0.0523, 0.0597], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0090, 0.0105, 0.0107, 0.0116, 0.0084, 0.0116, 0.0107], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:10:29,693 INFO [train.py:873] (1/4) Epoch 4, batch 4700, loss[loss=0.1818, simple_loss=0.1929, pruned_loss=0.08538, over 14311.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2006, pruned_loss=0.09141, over 1980618.98 frames. ], batch size: 31, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:10:56,345 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.544e+02 3.410e+02 4.417e+02 8.558e+02, threshold=6.819e+02, percent-clipped=1.0 2022-12-07 11:11:56,835 INFO [train.py:873] (1/4) Epoch 4, batch 4800, loss[loss=0.2028, simple_loss=0.1743, pruned_loss=0.1157, over 1277.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2002, pruned_loss=0.09139, over 1958338.12 frames. ], batch size: 100, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:12:11,709 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:12:13,270 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5357, 4.3181, 4.8165, 4.0578, 4.5639, 4.9823, 1.8190, 4.3386], device='cuda:1'), covar=tensor([0.0154, 0.0260, 0.0327, 0.0362, 0.0256, 0.0099, 0.2834, 0.0284], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0126, 0.0117, 0.0103, 0.0159, 0.0110, 0.0148, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 11:12:24,598 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 3.057e+02 3.716e+02 4.446e+02 1.010e+03, threshold=7.432e+02, percent-clipped=2.0 2022-12-07 11:12:54,318 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1639, 3.0609, 2.8434, 1.9365, 2.6810, 2.9585, 3.3381, 2.1177], device='cuda:1'), covar=tensor([0.0464, 0.2056, 0.0945, 0.2269, 0.0929, 0.0474, 0.0745, 0.2019], device='cuda:1'), in_proj_covar=tensor([0.0096, 0.0223, 0.0108, 0.0128, 0.0097, 0.0098, 0.0080, 0.0122], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:13:00,460 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:13:23,629 INFO [train.py:873] (1/4) Epoch 4, batch 4900, loss[loss=0.2025, simple_loss=0.216, pruned_loss=0.09447, over 14289.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2003, pruned_loss=0.09162, over 1973755.04 frames. ], batch size: 25, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:13:40,702 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:13:42,437 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:13:51,358 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.749e+02 3.689e+02 4.632e+02 8.661e+02, threshold=7.378e+02, percent-clipped=1.0 2022-12-07 11:13:52,117 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 11:13:55,049 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2725, 1.8688, 2.3228, 2.0378, 2.4835, 2.1346, 2.0528, 2.0846], device='cuda:1'), covar=tensor([0.0168, 0.0613, 0.0109, 0.0259, 0.0112, 0.0210, 0.0133, 0.0310], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0346, 0.0303, 0.0313, 0.0312, 0.0262, 0.0303, 0.0368], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:13:56,824 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9872, 2.1445, 4.0639, 4.1400, 4.2215, 2.6497, 4.3497, 3.1460], device='cuda:1'), covar=tensor([0.0078, 0.0272, 0.0210, 0.0089, 0.0057, 0.0356, 0.0040, 0.0247], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0177, 0.0247, 0.0202, 0.0160, 0.0224, 0.0138, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 11:13:59,476 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6201, 2.3640, 2.3331, 1.2433, 2.1603, 2.1379, 2.7763, 2.0148], device='cuda:1'), covar=tensor([0.0576, 0.1712, 0.1249, 0.3100, 0.1011, 0.0669, 0.0442, 0.1587], device='cuda:1'), in_proj_covar=tensor([0.0095, 0.0219, 0.0108, 0.0126, 0.0097, 0.0097, 0.0078, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:14:21,718 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:14:50,476 INFO [train.py:873] (1/4) Epoch 4, batch 5000, loss[loss=0.2486, simple_loss=0.2432, pruned_loss=0.127, over 10342.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2011, pruned_loss=0.09189, over 2018033.47 frames. ], batch size: 100, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:15:14,308 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:15:18,160 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.479e+02 3.081e+02 3.772e+02 7.331e+02, threshold=6.161e+02, percent-clipped=0.0 2022-12-07 11:16:06,376 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:16:16,333 INFO [train.py:873] (1/4) Epoch 4, batch 5100, loss[loss=0.1911, simple_loss=0.2034, pruned_loss=0.08939, over 14482.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2003, pruned_loss=0.09162, over 1984073.10 frames. ], batch size: 51, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:16:19,917 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 11:16:31,656 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:16:43,937 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.535e+02 3.343e+02 4.225e+02 1.167e+03, threshold=6.685e+02, percent-clipped=6.0 2022-12-07 11:17:02,943 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2022-12-07 11:17:12,659 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:17:43,049 INFO [train.py:873] (1/4) Epoch 4, batch 5200, loss[loss=0.2474, simple_loss=0.2274, pruned_loss=0.1337, over 8597.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2, pruned_loss=0.09162, over 1965269.11 frames. ], batch size: 100, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:18:10,408 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.745e+02 3.569e+02 4.514e+02 6.730e+02, threshold=7.138e+02, percent-clipped=2.0 2022-12-07 11:18:30,837 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6309, 0.5907, 0.7978, 0.4649, 0.6183, 0.7087, 0.7578, 0.5546], device='cuda:1'), covar=tensor([0.0234, 0.0160, 0.0122, 0.0101, 0.0389, 0.0323, 0.0086, 0.0251], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0011, 0.0012, 0.0015, 0.0012, 0.0018], device='cuda:1'), out_proj_covar=tensor([4.9683e-05, 4.8960e-05, 5.0580e-05, 4.7706e-05, 5.1360e-05, 6.2055e-05, 5.8566e-05, 7.0948e-05], device='cuda:1') 2022-12-07 11:18:38,860 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0098, 2.8486, 2.8157, 2.0216, 1.7170, 2.7578, 2.7107, 1.4320], device='cuda:1'), covar=tensor([0.3432, 0.0734, 0.1757, 0.1602, 0.1248, 0.0523, 0.1502, 0.3584], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0057, 0.0049, 0.0050, 0.0068, 0.0055, 0.0079, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:19:08,658 INFO [train.py:873] (1/4) Epoch 4, batch 5300, loss[loss=0.2288, simple_loss=0.2234, pruned_loss=0.1171, over 10369.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2001, pruned_loss=0.09184, over 1984071.68 frames. ], batch size: 100, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:19:36,271 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.605e+02 3.441e+02 4.511e+02 8.751e+02, threshold=6.883e+02, percent-clipped=1.0 2022-12-07 11:20:06,260 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9384, 3.3686, 4.2640, 2.9437, 2.8423, 3.1153, 1.4814, 2.9850], device='cuda:1'), covar=tensor([0.1429, 0.1245, 0.0832, 0.1804, 0.1740, 0.1517, 0.6233, 0.3020], device='cuda:1'), in_proj_covar=tensor([0.0070, 0.0076, 0.0069, 0.0083, 0.0096, 0.0065, 0.0140, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:20:12,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5384, 1.1384, 2.0567, 1.8216, 2.0290, 1.9678, 1.5164, 2.0598], device='cuda:1'), covar=tensor([0.0358, 0.0704, 0.0090, 0.0223, 0.0152, 0.0088, 0.0249, 0.0097], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0154, 0.0097, 0.0128, 0.0109, 0.0116, 0.0083, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:20:20,551 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:20:33,895 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 11:20:35,860 INFO [train.py:873] (1/4) Epoch 4, batch 5400, loss[loss=0.2018, simple_loss=0.2088, pruned_loss=0.0974, over 14264.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2007, pruned_loss=0.09265, over 1936878.27 frames. ], batch size: 57, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:20:48,002 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3086, 0.8066, 0.9144, 0.8948, 1.0154, 1.0274, 1.1415, 1.1739], device='cuda:1'), covar=tensor([0.0394, 0.0747, 0.0266, 0.2299, 0.0666, 0.0271, 0.0398, 0.0412], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0018, 0.0019, 0.0017, 0.0018, 0.0023, 0.0017, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.7982e-05, 6.4164e-05, 6.2558e-05, 6.2683e-05, 6.0250e-05, 7.7622e-05, 6.5001e-05, 6.1297e-05], device='cuda:1') 2022-12-07 11:21:03,785 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.371e+02 3.121e+02 4.059e+02 8.001e+02, threshold=6.242e+02, percent-clipped=5.0 2022-12-07 11:21:29,967 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3379, 0.8866, 0.8977, 1.0249, 1.2350, 0.8157, 1.2247, 1.2288], device='cuda:1'), covar=tensor([0.0350, 0.0601, 0.0295, 0.0484, 0.0408, 0.0285, 0.0295, 0.0535], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0018, 0.0019, 0.0017, 0.0017, 0.0023, 0.0016, 0.0018], device='cuda:1'), out_proj_covar=tensor([5.8455e-05, 6.3133e-05, 6.2123e-05, 6.1341e-05, 5.8483e-05, 7.7571e-05, 6.3931e-05, 6.1020e-05], device='cuda:1') 2022-12-07 11:22:02,364 INFO [train.py:873] (1/4) Epoch 4, batch 5500, loss[loss=0.1804, simple_loss=0.1937, pruned_loss=0.08351, over 14512.00 frames. ], tot_loss[loss=0.191, simple_loss=0.1997, pruned_loss=0.09117, over 1943754.39 frames. ], batch size: 51, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:22:03,983 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 11:22:30,325 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.632e+02 3.347e+02 4.419e+02 1.188e+03, threshold=6.694e+02, percent-clipped=11.0 2022-12-07 11:22:39,521 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:22:59,285 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9253, 1.3558, 3.0693, 1.3253, 3.1273, 3.0863, 2.2564, 3.1837], device='cuda:1'), covar=tensor([0.0186, 0.2524, 0.0232, 0.2059, 0.0252, 0.0275, 0.0549, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0158, 0.0128, 0.0167, 0.0143, 0.0135, 0.0115, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 11:23:01,935 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:23:29,473 INFO [train.py:873] (1/4) Epoch 4, batch 5600, loss[loss=0.1952, simple_loss=0.2051, pruned_loss=0.09271, over 14346.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2009, pruned_loss=0.09257, over 1965261.71 frames. ], batch size: 66, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:23:29,665 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:23:32,025 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:23:39,222 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 11:23:54,621 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:23:57,191 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.230e+02 2.669e+02 3.555e+02 4.575e+02 8.799e+02, threshold=7.110e+02, percent-clipped=5.0 2022-12-07 11:24:21,887 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:24:38,521 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8538, 2.6371, 2.6465, 2.8949, 2.5177, 2.3983, 2.8452, 2.8006], device='cuda:1'), covar=tensor([0.0638, 0.0699, 0.0685, 0.0574, 0.0783, 0.0770, 0.0697, 0.0779], device='cuda:1'), in_proj_covar=tensor([0.0105, 0.0086, 0.0104, 0.0104, 0.0111, 0.0083, 0.0114, 0.0106], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:24:41,074 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:24:44,083 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.50 vs. limit=2.0 2022-12-07 11:24:55,323 INFO [train.py:873] (1/4) Epoch 4, batch 5700, loss[loss=0.1898, simple_loss=0.1931, pruned_loss=0.09329, over 6920.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2001, pruned_loss=0.09159, over 1991120.08 frames. ], batch size: 100, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:25:12,264 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=17.39 vs. limit=5.0 2022-12-07 11:25:19,511 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8676, 4.4167, 4.4361, 4.9565, 4.5072, 4.0945, 4.9436, 4.8237], device='cuda:1'), covar=tensor([0.0691, 0.0641, 0.0544, 0.0579, 0.0570, 0.0525, 0.0576, 0.0711], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0087, 0.0105, 0.0105, 0.0113, 0.0084, 0.0116, 0.0106], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:25:22,827 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:25:23,563 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 2.835e+02 3.398e+02 4.455e+02 8.619e+02, threshold=6.796e+02, percent-clipped=2.0 2022-12-07 11:26:22,923 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:26:23,532 INFO [train.py:873] (1/4) Epoch 4, batch 5800, loss[loss=0.1867, simple_loss=0.1998, pruned_loss=0.08684, over 14232.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.1998, pruned_loss=0.09163, over 1962483.61 frames. ], batch size: 32, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:26:27,989 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:26:52,116 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.705e+02 3.370e+02 4.199e+02 7.683e+02, threshold=6.740e+02, percent-clipped=3.0 2022-12-07 11:27:16,488 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:21,793 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:27,096 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7275, 1.7113, 1.9655, 1.8210, 1.5051, 1.7739, 1.6267, 0.9977], device='cuda:1'), covar=tensor([0.2043, 0.0898, 0.1115, 0.0729, 0.1168, 0.0523, 0.1832, 0.3158], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0055, 0.0051, 0.0047, 0.0070, 0.0052, 0.0079, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:27:49,621 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:51,188 INFO [train.py:873] (1/4) Epoch 4, batch 5900, loss[loss=0.1847, simple_loss=0.1975, pruned_loss=0.08594, over 14289.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.1994, pruned_loss=0.09117, over 1943473.87 frames. ], batch size: 63, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:28:13,034 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:28:19,509 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.929e+02 3.624e+02 4.382e+02 1.301e+03, threshold=7.248e+02, percent-clipped=4.0 2022-12-07 11:28:40,751 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:28:58,591 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2282, 0.9501, 0.7301, 0.8796, 1.2206, 0.5527, 0.8658, 1.1499], device='cuda:1'), covar=tensor([0.0308, 0.0520, 0.0250, 0.0478, 0.0303, 0.0284, 0.0338, 0.0288], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0018, 0.0018, 0.0016, 0.0016, 0.0022, 0.0016, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.5487e-05, 6.1476e-05, 5.9930e-05, 5.9377e-05, 5.7203e-05, 7.5544e-05, 6.3686e-05, 5.9110e-05], device='cuda:1') 2022-12-07 11:29:19,378 INFO [train.py:873] (1/4) Epoch 4, batch 6000, loss[loss=0.1838, simple_loss=0.2012, pruned_loss=0.08314, over 14585.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.1993, pruned_loss=0.09085, over 1962620.27 frames. ], batch size: 22, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:29:19,378 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 11:29:30,470 INFO [train.py:905] (1/4) Epoch 4, validation: loss=0.1258, simple_loss=0.1688, pruned_loss=0.04138, over 857387.00 frames. 2022-12-07 11:29:30,471 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 11:29:55,365 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0649, 2.0816, 4.2471, 4.2574, 4.4830, 2.6154, 4.4366, 3.2426], device='cuda:1'), covar=tensor([0.0098, 0.0291, 0.0306, 0.0109, 0.0059, 0.0431, 0.0039, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0176, 0.0248, 0.0203, 0.0163, 0.0223, 0.0142, 0.0217], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 11:29:58,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.688e+02 3.368e+02 4.685e+02 1.249e+03, threshold=6.736e+02, percent-clipped=4.0 2022-12-07 11:30:22,376 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 11:30:56,777 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9043, 2.8005, 2.6473, 1.7351, 2.4046, 2.6713, 3.0599, 2.2105], device='cuda:1'), covar=tensor([0.0703, 0.2094, 0.1131, 0.2885, 0.1049, 0.0501, 0.0710, 0.2207], device='cuda:1'), in_proj_covar=tensor([0.0093, 0.0214, 0.0108, 0.0126, 0.0097, 0.0095, 0.0080, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:30:58,686 INFO [train.py:873] (1/4) Epoch 4, batch 6100, loss[loss=0.1535, simple_loss=0.1825, pruned_loss=0.06224, over 13925.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.1999, pruned_loss=0.09093, over 2023028.17 frames. ], batch size: 23, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:31:27,398 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.104e+02 2.717e+02 3.239e+02 4.102e+02 7.823e+02, threshold=6.479e+02, percent-clipped=2.0 2022-12-07 11:31:39,789 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6960, 0.5634, 0.7700, 0.4131, 0.5994, 0.5583, 0.4780, 0.5207], device='cuda:1'), covar=tensor([0.0198, 0.0121, 0.0224, 0.0121, 0.0320, 0.0352, 0.0179, 0.0412], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0013, 0.0012, 0.0013, 0.0016, 0.0012, 0.0018], device='cuda:1'), out_proj_covar=tensor([5.3240e-05, 5.3985e-05, 5.6828e-05, 5.1192e-05, 5.5006e-05, 6.6733e-05, 6.1961e-05, 7.3678e-05], device='cuda:1') 2022-12-07 11:31:47,359 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:31:52,827 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:31:54,402 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-07 11:32:04,579 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:32:07,987 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6239, 2.5595, 2.6592, 2.6257, 2.6374, 2.4740, 1.2352, 2.4040], device='cuda:1'), covar=tensor([0.0242, 0.0283, 0.0363, 0.0219, 0.0263, 0.0525, 0.2039, 0.0287], device='cuda:1'), in_proj_covar=tensor([0.0120, 0.0125, 0.0118, 0.0101, 0.0155, 0.0110, 0.0145, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 11:32:09,907 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6532, 2.0288, 3.6872, 3.8134, 3.6326, 2.2956, 3.7841, 2.9313], device='cuda:1'), covar=tensor([0.0086, 0.0274, 0.0193, 0.0095, 0.0070, 0.0389, 0.0055, 0.0272], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0177, 0.0253, 0.0205, 0.0164, 0.0227, 0.0147, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 11:32:25,365 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:32:27,179 INFO [train.py:873] (1/4) Epoch 4, batch 6200, loss[loss=0.1735, simple_loss=0.2018, pruned_loss=0.07257, over 14023.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.1998, pruned_loss=0.09059, over 1989789.86 frames. ], batch size: 22, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:32:39,444 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 11:32:48,603 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:32:55,224 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.759e+02 3.767e+02 5.172e+02 1.797e+03, threshold=7.535e+02, percent-clipped=14.0 2022-12-07 11:32:57,903 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:07,493 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:16,611 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:30,102 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:33:55,201 INFO [train.py:873] (1/4) Epoch 4, batch 6300, loss[loss=0.1858, simple_loss=0.1999, pruned_loss=0.08586, over 14269.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.1982, pruned_loss=0.08896, over 2010085.02 frames. ], batch size: 76, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:33:58,966 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:34:08,988 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:34:23,887 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.322e+02 3.000e+02 3.617e+02 7.085e+02, threshold=6.000e+02, percent-clipped=0.0 2022-12-07 11:35:03,257 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:35:14,662 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7791, 3.5456, 3.2803, 3.3236, 3.6440, 3.6065, 3.7367, 3.6400], device='cuda:1'), covar=tensor([0.0729, 0.0582, 0.1649, 0.2689, 0.0693, 0.0703, 0.1068, 0.1075], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0216, 0.0326, 0.0416, 0.0248, 0.0289, 0.0297, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:35:24,089 INFO [train.py:873] (1/4) Epoch 4, batch 6400, loss[loss=0.152, simple_loss=0.1758, pruned_loss=0.06406, over 13883.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.199, pruned_loss=0.08943, over 2025952.34 frames. ], batch size: 23, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:35:41,052 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5611, 4.9246, 5.0661, 5.4413, 5.1187, 4.6171, 5.4456, 4.5547], device='cuda:1'), covar=tensor([0.0219, 0.0778, 0.0227, 0.0352, 0.0580, 0.0283, 0.0461, 0.0376], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0201, 0.0134, 0.0127, 0.0133, 0.0107, 0.0194, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:35:52,467 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.620e+02 3.469e+02 4.373e+02 1.002e+03, threshold=6.937e+02, percent-clipped=9.0 2022-12-07 11:36:12,728 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:36:17,884 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:36:28,569 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0752, 1.9878, 4.2023, 3.9747, 3.9496, 4.2451, 3.7203, 4.2793], device='cuda:1'), covar=tensor([0.1073, 0.1101, 0.0080, 0.0102, 0.0130, 0.0082, 0.0157, 0.0095], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0150, 0.0095, 0.0128, 0.0109, 0.0115, 0.0082, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:36:52,305 INFO [train.py:873] (1/4) Epoch 4, batch 6500, loss[loss=0.1984, simple_loss=0.2121, pruned_loss=0.09232, over 14237.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.199, pruned_loss=0.08975, over 1987979.33 frames. ], batch size: 39, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:36:54,888 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:00,023 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:08,986 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8652, 0.9198, 0.7008, 0.9454, 0.9925, 0.6360, 0.9550, 1.1061], device='cuda:1'), covar=tensor([0.0449, 0.0405, 0.0335, 0.0356, 0.0235, 0.0312, 0.0265, 0.0182], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0020, 0.0020, 0.0017, 0.0018, 0.0024, 0.0018, 0.0018], device='cuda:1'), out_proj_covar=tensor([6.4951e-05, 6.9281e-05, 6.5276e-05, 6.4698e-05, 6.3442e-05, 8.2790e-05, 6.9223e-05, 6.2965e-05], device='cuda:1') 2022-12-07 11:37:18,786 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:20,462 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.481e+02 2.712e+02 3.572e+02 4.670e+02 1.139e+03, threshold=7.145e+02, percent-clipped=7.0 2022-12-07 11:37:35,579 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:35,962 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-07 11:37:43,058 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 11:38:12,751 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8635, 1.5458, 2.0629, 1.7126, 1.9291, 1.3248, 1.6279, 1.7137], device='cuda:1'), covar=tensor([0.1044, 0.3114, 0.0237, 0.2122, 0.0365, 0.1541, 0.1357, 0.0476], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0283, 0.0172, 0.0382, 0.0180, 0.0289, 0.0267, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:38:20,130 INFO [train.py:873] (1/4) Epoch 4, batch 6600, loss[loss=0.1717, simple_loss=0.1761, pruned_loss=0.0837, over 3852.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.1988, pruned_loss=0.08994, over 1941007.88 frames. ], batch size: 100, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:38:29,414 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:38:48,062 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.617e+02 3.366e+02 4.262e+02 9.069e+02, threshold=6.731e+02, percent-clipped=3.0 2022-12-07 11:39:22,480 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:39:37,196 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=3.68 vs. limit=2.0 2022-12-07 11:39:47,260 INFO [train.py:873] (1/4) Epoch 4, batch 6700, loss[loss=0.1902, simple_loss=0.2069, pruned_loss=0.0868, over 14323.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.1995, pruned_loss=0.09077, over 1947969.48 frames. ], batch size: 31, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:40:14,795 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.696e+02 3.507e+02 4.703e+02 1.037e+03, threshold=7.015e+02, percent-clipped=5.0 2022-12-07 11:41:06,297 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 11:41:13,497 INFO [train.py:873] (1/4) Epoch 4, batch 6800, loss[loss=0.1781, simple_loss=0.1955, pruned_loss=0.08033, over 14264.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.199, pruned_loss=0.0909, over 1929944.36 frames. ], batch size: 57, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:41:39,813 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:41:42,140 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.836e+02 3.618e+02 4.474e+02 6.792e+02, threshold=7.235e+02, percent-clipped=0.0 2022-12-07 11:42:09,579 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:42:21,759 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:42:41,256 INFO [train.py:873] (1/4) Epoch 4, batch 6900, loss[loss=0.1741, simple_loss=0.1904, pruned_loss=0.07891, over 14019.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.1988, pruned_loss=0.09019, over 2001734.62 frames. ], batch size: 20, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:42:45,767 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:02,679 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:09,418 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.737e+02 3.624e+02 4.962e+02 1.279e+03, threshold=7.248e+02, percent-clipped=6.0 2022-12-07 11:43:18,588 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.34 vs. limit=5.0 2022-12-07 11:43:42,714 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:45,826 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:44:07,055 INFO [train.py:873] (1/4) Epoch 4, batch 7000, loss[loss=0.1471, simple_loss=0.1763, pruned_loss=0.05896, over 14045.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.1983, pruned_loss=0.09018, over 1944807.56 frames. ], batch size: 19, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:44:23,776 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:44:35,728 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.834e+02 3.387e+02 4.469e+02 7.718e+02, threshold=6.775e+02, percent-clipped=1.0 2022-12-07 11:44:38,278 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 11:45:33,333 INFO [train.py:873] (1/4) Epoch 4, batch 7100, loss[loss=0.2184, simple_loss=0.2076, pruned_loss=0.1146, over 7774.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.1987, pruned_loss=0.08942, over 1999440.81 frames. ], batch size: 100, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:45:46,856 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:45:51,512 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 11:46:02,683 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.583e+02 3.436e+02 4.712e+02 1.166e+03, threshold=6.872e+02, percent-clipped=4.0 2022-12-07 11:46:40,706 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:01,604 INFO [train.py:873] (1/4) Epoch 4, batch 7200, loss[loss=0.1817, simple_loss=0.1971, pruned_loss=0.08313, over 13871.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.1989, pruned_loss=0.08977, over 1980117.11 frames. ], batch size: 23, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:47:06,091 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:19,539 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:30,663 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.626e+02 3.344e+02 4.272e+02 1.259e+03, threshold=6.687e+02, percent-clipped=4.0 2022-12-07 11:47:47,983 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:55,030 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9429, 3.3995, 4.5303, 3.5121, 4.6434, 4.7237, 4.2907, 4.0290], device='cuda:1'), covar=tensor([0.0174, 0.1613, 0.0380, 0.0925, 0.0358, 0.0260, 0.1212, 0.1171], device='cuda:1'), in_proj_covar=tensor([0.0260, 0.0342, 0.0330, 0.0309, 0.0315, 0.0266, 0.0310, 0.0366], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:48:05,772 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1836, 1.0715, 0.7939, 1.1463, 1.1923, 1.1959, 1.1933, 1.2813], device='cuda:1'), covar=tensor([0.1122, 0.2504, 0.2579, 0.0895, 0.2269, 0.0564, 0.0354, 0.1233], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0010, 0.0011, 0.0015, 0.0012, 0.0016], device='cuda:1'), out_proj_covar=tensor([5.0767e-05, 5.2933e-05, 5.3121e-05, 4.7860e-05, 5.2270e-05, 6.6012e-05, 5.9536e-05, 6.8293e-05], device='cuda:1') 2022-12-07 11:48:10,816 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2048, 2.3513, 4.2337, 3.0308, 4.0869, 1.8972, 3.3671, 3.9390], device='cuda:1'), covar=tensor([0.0282, 0.4798, 0.0381, 0.8643, 0.0260, 0.3996, 0.1106, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0274, 0.0166, 0.0367, 0.0174, 0.0278, 0.0256, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:48:27,250 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1183, 2.0823, 1.8188, 1.8393, 2.0095, 2.0369, 2.0552, 2.0535], device='cuda:1'), covar=tensor([0.0689, 0.0851, 0.1507, 0.2122, 0.0989, 0.0951, 0.1396, 0.0878], device='cuda:1'), in_proj_covar=tensor([0.0261, 0.0226, 0.0332, 0.0429, 0.0256, 0.0297, 0.0312, 0.0252], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:48:28,907 INFO [train.py:873] (1/4) Epoch 4, batch 7300, loss[loss=0.1678, simple_loss=0.1871, pruned_loss=0.07425, over 14400.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.1976, pruned_loss=0.08881, over 1968249.79 frames. ], batch size: 41, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:48:59,514 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:49:01,143 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.428e+02 2.508e+02 3.092e+02 3.972e+02 1.080e+03, threshold=6.184e+02, percent-clipped=3.0 2022-12-07 11:49:16,081 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5000, 3.2265, 3.0438, 3.4662, 3.3981, 3.4021, 3.5540, 2.9082], device='cuda:1'), covar=tensor([0.0351, 0.0998, 0.0447, 0.0435, 0.0582, 0.0351, 0.0556, 0.0541], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0209, 0.0140, 0.0131, 0.0137, 0.0111, 0.0198, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:49:26,315 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5025, 1.9686, 3.6335, 2.4456, 3.4540, 1.7763, 2.6957, 3.2431], device='cuda:1'), covar=tensor([0.0484, 0.5564, 0.0373, 0.9830, 0.0321, 0.4180, 0.1524, 0.0408], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0268, 0.0162, 0.0361, 0.0170, 0.0269, 0.0251, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:49:59,921 INFO [train.py:873] (1/4) Epoch 4, batch 7400, loss[loss=0.1952, simple_loss=0.2047, pruned_loss=0.09282, over 14264.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.1985, pruned_loss=0.09001, over 1984796.69 frames. ], batch size: 76, lr: 1.76e-02, grad_scale: 8.0 2022-12-07 11:50:29,607 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.725e+02 3.785e+02 4.899e+02 8.566e+02, threshold=7.571e+02, percent-clipped=11.0 2022-12-07 11:50:33,833 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:50:34,343 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.59 vs. limit=2.0 2022-12-07 11:50:39,700 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9648, 0.8482, 0.8234, 1.0808, 1.0473, 0.6325, 1.1032, 0.9426], device='cuda:1'), covar=tensor([0.0530, 0.0715, 0.0341, 0.0363, 0.0257, 0.0336, 0.0352, 0.0348], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0019, 0.0017, 0.0018, 0.0024, 0.0019, 0.0019], device='cuda:1'), out_proj_covar=tensor([6.7083e-05, 6.9407e-05, 6.4767e-05, 6.4071e-05, 6.5947e-05, 8.5605e-05, 7.3185e-05, 6.7458e-05], device='cuda:1') 2022-12-07 11:51:02,852 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:27,271 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:28,010 INFO [train.py:873] (1/4) Epoch 4, batch 7500, loss[loss=0.1752, simple_loss=0.1554, pruned_loss=0.09754, over 1165.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.1989, pruned_loss=0.09073, over 1935638.16 frames. ], batch size: 100, lr: 1.76e-02, grad_scale: 8.0 2022-12-07 11:51:42,266 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 11:51:45,980 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:47,657 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:49,313 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9265, 2.1618, 3.9266, 2.8060, 3.8964, 2.0847, 2.8896, 3.7559], device='cuda:1'), covar=tensor([0.0390, 0.4901, 0.0378, 0.7934, 0.0250, 0.3913, 0.1285, 0.0194], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0276, 0.0165, 0.0367, 0.0175, 0.0278, 0.0256, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 11:51:56,539 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.609e+02 3.309e+02 4.020e+02 7.663e+02, threshold=6.618e+02, percent-clipped=1.0 2022-12-07 11:52:55,167 INFO [train.py:873] (1/4) Epoch 5, batch 0, loss[loss=0.2612, simple_loss=0.2541, pruned_loss=0.1342, over 14364.00 frames. ], tot_loss[loss=0.2612, simple_loss=0.2541, pruned_loss=0.1342, over 14364.00 frames. ], batch size: 73, lr: 1.64e-02, grad_scale: 8.0 2022-12-07 11:52:55,167 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 11:53:02,316 INFO [train.py:905] (1/4) Epoch 5, validation: loss=0.1366, simple_loss=0.1807, pruned_loss=0.04626, over 857387.00 frames. 2022-12-07 11:53:02,317 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 11:53:07,518 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:53:12,607 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2022-12-07 11:53:21,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0653, 1.7957, 4.1836, 4.0402, 3.9880, 4.3364, 3.9398, 4.3410], device='cuda:1'), covar=tensor([0.1158, 0.1124, 0.0074, 0.0114, 0.0127, 0.0082, 0.0100, 0.0072], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0150, 0.0095, 0.0130, 0.0110, 0.0114, 0.0083, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:53:21,389 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:54:03,555 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:54:05,438 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.465e+01 2.449e+02 3.602e+02 5.226e+02 1.817e+03, threshold=7.204e+02, percent-clipped=11.0 2022-12-07 11:54:30,453 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 11:54:31,490 INFO [train.py:873] (1/4) Epoch 5, batch 100, loss[loss=0.213, simple_loss=0.2158, pruned_loss=0.1051, over 14281.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2003, pruned_loss=0.08993, over 856398.80 frames. ], batch size: 63, lr: 1.64e-02, grad_scale: 8.0 2022-12-07 11:54:40,826 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:54:45,804 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:55:32,795 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.778e+02 3.417e+02 4.208e+02 9.792e+02, threshold=6.833e+02, percent-clipped=3.0 2022-12-07 11:55:33,883 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:55:34,399 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 11:55:58,008 INFO [train.py:873] (1/4) Epoch 5, batch 200, loss[loss=0.1652, simple_loss=0.18, pruned_loss=0.07523, over 13957.00 frames. ], tot_loss[loss=0.188, simple_loss=0.1985, pruned_loss=0.08876, over 1381758.56 frames. ], batch size: 19, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:56:06,135 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:56:11,370 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0290, 4.4271, 4.5124, 4.9541, 4.7489, 4.3021, 4.9059, 4.0345], device='cuda:1'), covar=tensor([0.0281, 0.1124, 0.0331, 0.0410, 0.0643, 0.0425, 0.0583, 0.0541], device='cuda:1'), in_proj_covar=tensor([0.0120, 0.0204, 0.0139, 0.0127, 0.0134, 0.0107, 0.0196, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 11:56:18,484 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4515, 1.1326, 1.3828, 0.8918, 1.1149, 1.3437, 1.2558, 1.2956], device='cuda:1'), covar=tensor([0.0247, 0.0928, 0.0417, 0.0411, 0.0586, 0.0373, 0.0294, 0.0650], device='cuda:1'), in_proj_covar=tensor([0.0097, 0.0220, 0.0115, 0.0127, 0.0097, 0.0095, 0.0083, 0.0122], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 11:56:26,423 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:56:48,243 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:57:00,756 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.564e+02 3.377e+02 4.486e+02 9.729e+02, threshold=6.754e+02, percent-clipped=3.0 2022-12-07 11:57:25,909 INFO [train.py:873] (1/4) Epoch 5, batch 300, loss[loss=0.145, simple_loss=0.172, pruned_loss=0.05902, over 13919.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.1962, pruned_loss=0.08712, over 1588725.48 frames. ], batch size: 19, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:57:39,894 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:58:19,684 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 11:58:27,947 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.626e+02 3.332e+02 4.486e+02 9.530e+02, threshold=6.664e+02, percent-clipped=3.0 2022-12-07 11:58:53,706 INFO [train.py:873] (1/4) Epoch 5, batch 400, loss[loss=0.2001, simple_loss=0.2027, pruned_loss=0.0988, over 8604.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.197, pruned_loss=0.08825, over 1734310.71 frames. ], batch size: 100, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:58:59,012 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:59:03,369 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 11:59:14,485 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1472, 1.8953, 4.3211, 4.2038, 4.0749, 4.4303, 4.4201, 4.2938], device='cuda:1'), covar=tensor([0.1127, 0.1281, 0.0142, 0.0120, 0.0150, 0.0116, 0.0062, 0.0179], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0156, 0.0100, 0.0134, 0.0115, 0.0119, 0.0086, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 11:59:52,580 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:59:52,650 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 11:59:56,042 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.557e+02 3.010e+02 3.908e+02 7.694e+02, threshold=6.020e+02, percent-clipped=1.0 2022-12-07 12:00:21,154 INFO [train.py:873] (1/4) Epoch 5, batch 500, loss[loss=0.158, simple_loss=0.1491, pruned_loss=0.08343, over 2646.00 frames. ], tot_loss[loss=0.185, simple_loss=0.1962, pruned_loss=0.0869, over 1868289.48 frames. ], batch size: 100, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 12:00:30,480 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.17 vs. limit=5.0 2022-12-07 12:00:49,042 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:01:23,929 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.567e+02 3.338e+02 3.998e+02 8.568e+02, threshold=6.676e+02, percent-clipped=3.0 2022-12-07 12:01:31,361 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:01:48,857 INFO [train.py:873] (1/4) Epoch 5, batch 600, loss[loss=0.2107, simple_loss=0.2092, pruned_loss=0.1061, over 4987.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.1956, pruned_loss=0.08664, over 1912973.39 frames. ], batch size: 100, lr: 1.62e-02, grad_scale: 4.0 2022-12-07 12:02:02,954 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:26,755 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1233, 1.9707, 3.3196, 2.4246, 3.1599, 1.9274, 2.5187, 2.9544], device='cuda:1'), covar=tensor([0.0759, 0.4894, 0.0390, 0.6699, 0.0387, 0.3357, 0.1374, 0.0471], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0274, 0.0164, 0.0365, 0.0176, 0.0274, 0.0253, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:02:36,310 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:41,518 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.29 vs. limit=2.0 2022-12-07 12:02:45,023 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:51,900 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.564e+02 3.109e+02 3.998e+02 8.746e+02, threshold=6.217e+02, percent-clipped=4.0 2022-12-07 12:02:55,527 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:03,353 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:16,253 INFO [train.py:873] (1/4) Epoch 5, batch 700, loss[loss=0.1629, simple_loss=0.1881, pruned_loss=0.0689, over 14561.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.1953, pruned_loss=0.0861, over 2014151.85 frames. ], batch size: 34, lr: 1.62e-02, grad_scale: 4.0 2022-12-07 12:03:21,155 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.25 vs. limit=5.0 2022-12-07 12:03:29,753 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:49,031 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:56,509 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:04:10,434 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 12:04:10,492 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:04:11,815 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2022-12-07 12:04:14,766 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:04:18,898 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.586e+02 3.335e+02 4.382e+02 1.879e+03, threshold=6.670e+02, percent-clipped=6.0 2022-12-07 12:04:43,827 INFO [train.py:873] (1/4) Epoch 5, batch 800, loss[loss=0.228, simple_loss=0.2107, pruned_loss=0.1227, over 3897.00 frames. ], tot_loss[loss=0.184, simple_loss=0.195, pruned_loss=0.08653, over 2000285.80 frames. ], batch size: 100, lr: 1.62e-02, grad_scale: 8.0 2022-12-07 12:04:57,145 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:05:04,572 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:05:24,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0645, 2.5655, 3.7202, 2.7395, 3.6271, 3.7903, 3.3888, 3.0641], device='cuda:1'), covar=tensor([0.0282, 0.2172, 0.0573, 0.1438, 0.0630, 0.0427, 0.1559, 0.1474], device='cuda:1'), in_proj_covar=tensor([0.0262, 0.0338, 0.0327, 0.0308, 0.0322, 0.0268, 0.0314, 0.0359], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:05:29,989 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=3.04 vs. limit=2.0 2022-12-07 12:05:47,140 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.770e+02 3.066e+02 3.847e+02 6.961e+02, threshold=6.133e+02, percent-clipped=2.0 2022-12-07 12:06:06,958 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 12:06:11,657 INFO [train.py:873] (1/4) Epoch 5, batch 900, loss[loss=0.2169, simple_loss=0.1917, pruned_loss=0.121, over 2574.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.1943, pruned_loss=0.08549, over 1997547.76 frames. ], batch size: 100, lr: 1.62e-02, grad_scale: 8.0 2022-12-07 12:07:14,864 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.339e+02 3.126e+02 3.777e+02 6.609e+02, threshold=6.252e+02, percent-clipped=2.0 2022-12-07 12:07:24,722 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4039, 1.8595, 2.5031, 2.6622, 2.3703, 1.8295, 2.6579, 2.0836], device='cuda:1'), covar=tensor([0.0087, 0.0222, 0.0119, 0.0075, 0.0091, 0.0302, 0.0068, 0.0204], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0184, 0.0264, 0.0208, 0.0167, 0.0232, 0.0156, 0.0223], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:07:34,524 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-07 12:07:39,742 INFO [train.py:873] (1/4) Epoch 5, batch 1000, loss[loss=0.2139, simple_loss=0.1824, pruned_loss=0.1227, over 1266.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.1937, pruned_loss=0.08497, over 1994693.81 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:07:48,303 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:03,555 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2022-12-07 12:08:07,826 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:15,233 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:33,850 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:41,631 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9553, 2.9695, 3.1138, 3.0069, 3.0403, 2.9497, 1.3516, 2.8718], device='cuda:1'), covar=tensor([0.0247, 0.0277, 0.0377, 0.0294, 0.0251, 0.0483, 0.2679, 0.0270], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0131, 0.0119, 0.0105, 0.0163, 0.0115, 0.0150, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:08:42,296 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.699e+02 3.594e+02 4.260e+02 1.045e+03, threshold=7.189e+02, percent-clipped=6.0 2022-12-07 12:09:06,867 INFO [train.py:873] (1/4) Epoch 5, batch 1100, loss[loss=0.256, simple_loss=0.235, pruned_loss=0.1385, over 7765.00 frames. ], tot_loss[loss=0.182, simple_loss=0.1941, pruned_loss=0.08499, over 2047615.59 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:09:15,219 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:09:22,322 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:09:49,686 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7669, 1.9525, 2.2478, 1.4548, 1.5959, 1.9699, 1.3001, 1.8891], device='cuda:1'), covar=tensor([0.1327, 0.1463, 0.0475, 0.2468, 0.2512, 0.0757, 0.4128, 0.0892], device='cuda:1'), in_proj_covar=tensor([0.0069, 0.0073, 0.0069, 0.0083, 0.0098, 0.0064, 0.0135, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 12:09:59,538 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:10:09,480 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.558e+02 3.355e+02 4.456e+02 1.089e+03, threshold=6.709e+02, percent-clipped=2.0 2022-12-07 12:10:10,171 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-07 12:10:12,404 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8957, 2.1358, 2.8565, 2.4208, 2.9547, 2.7340, 2.6327, 2.3991], device='cuda:1'), covar=tensor([0.0250, 0.1935, 0.0498, 0.1090, 0.0459, 0.0455, 0.0619, 0.1685], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0340, 0.0338, 0.0312, 0.0331, 0.0271, 0.0321, 0.0360], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:10:34,137 INFO [train.py:873] (1/4) Epoch 5, batch 1200, loss[loss=0.2529, simple_loss=0.219, pruned_loss=0.1434, over 3862.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.1948, pruned_loss=0.0862, over 1987070.06 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:10:51,765 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2309, 1.8194, 2.2532, 2.5059, 1.9720, 1.8087, 2.2464, 2.1261], device='cuda:1'), covar=tensor([0.0046, 0.0111, 0.0056, 0.0032, 0.0062, 0.0170, 0.0064, 0.0077], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0188, 0.0269, 0.0213, 0.0170, 0.0237, 0.0157, 0.0230], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:10:52,731 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:11:03,998 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0080, 1.9578, 2.0171, 2.1116, 1.9937, 1.7547, 1.2007, 1.7508], device='cuda:1'), covar=tensor([0.0286, 0.0302, 0.0419, 0.0231, 0.0299, 0.0815, 0.1660, 0.0331], device='cuda:1'), in_proj_covar=tensor([0.0124, 0.0129, 0.0121, 0.0105, 0.0163, 0.0113, 0.0149, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:11:10,728 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:11:36,742 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4569, 1.2619, 3.6329, 1.4152, 3.4977, 3.6199, 2.5082, 3.8002], device='cuda:1'), covar=tensor([0.0210, 0.2922, 0.0281, 0.2320, 0.0450, 0.0275, 0.0824, 0.0147], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0156, 0.0130, 0.0165, 0.0149, 0.0135, 0.0114, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:11:37,430 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.031e+02 2.496e+02 3.332e+02 4.341e+02 8.542e+02, threshold=6.664e+02, percent-clipped=7.0 2022-12-07 12:11:44,180 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2022-12-07 12:12:02,328 INFO [train.py:873] (1/4) Epoch 5, batch 1300, loss[loss=0.2468, simple_loss=0.2171, pruned_loss=0.1383, over 4993.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.1938, pruned_loss=0.08566, over 1949493.64 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:12:05,338 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:12:11,352 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:15,127 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1374, 4.8445, 4.5942, 5.1059, 4.7329, 4.1486, 5.1147, 5.0431], device='cuda:1'), covar=tensor([0.0512, 0.0453, 0.0535, 0.0461, 0.0539, 0.0439, 0.0543, 0.0587], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0088, 0.0106, 0.0105, 0.0114, 0.0084, 0.0121, 0.0105], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:12:30,321 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:38,826 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:54,065 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:05,507 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.035e+02 2.616e+02 3.313e+02 3.855e+02 7.867e+02, threshold=6.626e+02, percent-clipped=4.0 2022-12-07 12:13:13,017 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:20,931 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:21,901 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7857, 3.3226, 2.4525, 3.8926, 3.7146, 3.7429, 2.9751, 2.5144], device='cuda:1'), covar=tensor([0.0447, 0.1222, 0.4602, 0.0265, 0.0555, 0.1340, 0.1280, 0.4831], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0300, 0.0317, 0.0184, 0.0240, 0.0239, 0.0259, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:13:30,390 INFO [train.py:873] (1/4) Epoch 5, batch 1400, loss[loss=0.1851, simple_loss=0.1604, pruned_loss=0.1049, over 1236.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.195, pruned_loss=0.08679, over 1974521.55 frames. ], batch size: 100, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:13:33,845 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5044, 3.3034, 2.6162, 1.7874, 3.0167, 2.9903, 3.5761, 2.5064], device='cuda:1'), covar=tensor([0.0576, 0.3148, 0.1534, 0.2910, 0.0550, 0.0678, 0.0789, 0.1842], device='cuda:1'), in_proj_covar=tensor([0.0098, 0.0216, 0.0117, 0.0129, 0.0097, 0.0100, 0.0083, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:13:35,482 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9707, 0.9511, 0.9603, 1.0384, 0.5331, 0.7210, 0.6277, 0.6340], device='cuda:1'), covar=tensor([0.0178, 0.0082, 0.0134, 0.0140, 0.0413, 0.0312, 0.0181, 0.0457], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0011, 0.0011, 0.0010, 0.0012, 0.0014, 0.0012, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.4968e-05, 5.4572e-05, 5.4580e-05, 4.9194e-05, 5.6815e-05, 6.9345e-05, 6.3667e-05, 7.5416e-05], device='cuda:1') 2022-12-07 12:13:45,805 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:14:13,994 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6556, 0.4909, 0.5489, 0.6727, 0.7314, 0.0882, 0.4639, 0.5225], device='cuda:1'), covar=tensor([0.0132, 0.0185, 0.0125, 0.0184, 0.0116, 0.0096, 0.0517, 0.0071], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0018, 0.0016, 0.0018, 0.0024, 0.0019, 0.0018], device='cuda:1'), out_proj_covar=tensor([6.6581e-05, 6.7960e-05, 6.2413e-05, 6.4174e-05, 6.6957e-05, 8.8279e-05, 7.2862e-05, 6.7266e-05], device='cuda:1') 2022-12-07 12:14:18,243 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:14:28,097 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:14:33,108 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.505e+02 2.718e+02 3.448e+02 4.143e+02 7.580e+02, threshold=6.896e+02, percent-clipped=4.0 2022-12-07 12:14:57,262 INFO [train.py:873] (1/4) Epoch 5, batch 1500, loss[loss=0.1952, simple_loss=0.1772, pruned_loss=0.1066, over 2650.00 frames. ], tot_loss[loss=0.183, simple_loss=0.1945, pruned_loss=0.08576, over 2031482.38 frames. ], batch size: 100, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:15:11,596 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:15:11,674 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:15:19,154 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-07 12:15:31,498 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:00,992 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.523e+02 3.162e+02 3.978e+02 7.805e+02, threshold=6.325e+02, percent-clipped=2.0 2022-12-07 12:16:02,931 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:24,830 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:16:26,441 INFO [train.py:873] (1/4) Epoch 5, batch 1600, loss[loss=0.1836, simple_loss=0.1945, pruned_loss=0.08634, over 14284.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.195, pruned_loss=0.08616, over 2043542.55 frames. ], batch size: 39, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:16:26,616 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:30,053 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9733, 2.0126, 1.9198, 2.0557, 1.7156, 1.8082, 2.0033, 2.0322], device='cuda:1'), covar=tensor([0.0898, 0.0785, 0.0859, 0.0817, 0.1221, 0.0860, 0.1038, 0.0937], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0089, 0.0108, 0.0105, 0.0115, 0.0086, 0.0121, 0.0108], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:16:52,367 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2022-12-07 12:16:52,814 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:58,365 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:25,470 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:30,354 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.483e+02 3.139e+02 4.013e+02 6.840e+02, threshold=6.277e+02, percent-clipped=1.0 2022-12-07 12:17:46,239 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:54,774 INFO [train.py:873] (1/4) Epoch 5, batch 1700, loss[loss=0.1475, simple_loss=0.1658, pruned_loss=0.06459, over 13664.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.1946, pruned_loss=0.08535, over 2070670.58 frames. ], batch size: 17, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:17:56,112 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:13,582 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4387, 1.3474, 3.5103, 1.3550, 3.3372, 3.4965, 2.6106, 3.7076], device='cuda:1'), covar=tensor([0.0204, 0.2730, 0.0283, 0.2374, 0.0502, 0.0284, 0.0624, 0.0154], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0159, 0.0131, 0.0169, 0.0150, 0.0139, 0.0118, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:18:18,867 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:50,914 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:59,553 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.546e+02 3.116e+02 4.047e+02 1.395e+03, threshold=6.231e+02, percent-clipped=6.0 2022-12-07 12:19:13,792 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1847, 2.4194, 4.0544, 4.2116, 4.2147, 2.8699, 4.2847, 3.3751], device='cuda:1'), covar=tensor([0.0100, 0.0321, 0.0457, 0.0129, 0.0085, 0.0431, 0.0076, 0.0347], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0188, 0.0268, 0.0212, 0.0172, 0.0234, 0.0156, 0.0225], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:19:25,247 INFO [train.py:873] (1/4) Epoch 5, batch 1800, loss[loss=0.1702, simple_loss=0.1461, pruned_loss=0.09717, over 2631.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.194, pruned_loss=0.08565, over 1954391.94 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:19:34,869 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:19:39,347 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:13,344 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 12:20:22,880 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:30,237 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.694e+02 3.195e+02 4.070e+02 6.631e+02, threshold=6.390e+02, percent-clipped=2.0 2022-12-07 12:20:51,093 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:53,873 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:20:55,470 INFO [train.py:873] (1/4) Epoch 5, batch 1900, loss[loss=0.1676, simple_loss=0.1899, pruned_loss=0.07259, over 14036.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.1935, pruned_loss=0.08489, over 2007480.54 frames. ], batch size: 26, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:21:23,359 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:21:38,188 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:22:01,271 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.008e+02 2.535e+02 3.167e+02 4.268e+02 9.066e+02, threshold=6.334e+02, percent-clipped=2.0 2022-12-07 12:22:13,206 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:22:26,466 INFO [train.py:873] (1/4) Epoch 5, batch 2000, loss[loss=0.1838, simple_loss=0.1571, pruned_loss=0.1052, over 1229.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.1937, pruned_loss=0.08539, over 2015453.49 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:22:34,555 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1084, 2.1382, 3.1047, 3.2140, 3.1724, 2.1064, 3.1776, 2.5545], device='cuda:1'), covar=tensor([0.0115, 0.0295, 0.0316, 0.0138, 0.0089, 0.0442, 0.0081, 0.0331], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0189, 0.0270, 0.0211, 0.0173, 0.0234, 0.0156, 0.0225], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:22:45,447 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:22:50,343 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0249, 1.9320, 2.0030, 2.0765, 1.9698, 1.9154, 1.1697, 1.7551], device='cuda:1'), covar=tensor([0.0296, 0.0369, 0.0457, 0.0228, 0.0321, 0.0532, 0.1695, 0.0363], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0136, 0.0126, 0.0109, 0.0167, 0.0114, 0.0152, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:23:16,446 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:23:30,258 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.510e+02 3.096e+02 3.855e+02 9.073e+02, threshold=6.191e+02, percent-clipped=1.0 2022-12-07 12:23:54,815 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 12:23:55,267 INFO [train.py:873] (1/4) Epoch 5, batch 2100, loss[loss=0.2179, simple_loss=0.2123, pruned_loss=0.1118, over 8622.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.1931, pruned_loss=0.08455, over 2009941.97 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:24:05,683 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:24:17,728 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.26 vs. limit=5.0 2022-12-07 12:24:50,030 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:24:58,325 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4322, 4.2700, 4.4543, 3.8317, 4.3466, 4.5793, 1.7074, 4.1969], device='cuda:1'), covar=tensor([0.0138, 0.0196, 0.0402, 0.0493, 0.0268, 0.0167, 0.2916, 0.0203], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0134, 0.0128, 0.0110, 0.0167, 0.0116, 0.0151, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:25:01,675 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.695e+02 3.113e+02 4.251e+02 1.070e+03, threshold=6.225e+02, percent-clipped=3.0 2022-12-07 12:25:23,313 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:25:27,727 INFO [train.py:873] (1/4) Epoch 5, batch 2200, loss[loss=0.1879, simple_loss=0.1853, pruned_loss=0.0952, over 4963.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.1938, pruned_loss=0.08548, over 1999891.53 frames. ], batch size: 100, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:25:54,694 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:03,687 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-07 12:26:05,773 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:22,398 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2446, 1.7169, 2.1511, 2.0188, 2.2310, 2.0696, 1.9659, 2.0044], device='cuda:1'), covar=tensor([0.0271, 0.1263, 0.0269, 0.0567, 0.0182, 0.0327, 0.0232, 0.0550], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0355, 0.0349, 0.0328, 0.0348, 0.0284, 0.0333, 0.0374], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:26:31,377 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.575e+02 3.323e+02 4.480e+02 8.616e+02, threshold=6.645e+02, percent-clipped=4.0 2022-12-07 12:26:36,966 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:43,164 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:56,111 INFO [train.py:873] (1/4) Epoch 5, batch 2300, loss[loss=0.1583, simple_loss=0.1845, pruned_loss=0.06604, over 14287.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.1929, pruned_loss=0.08499, over 1966745.49 frames. ], batch size: 44, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:27:15,933 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:25,526 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:48,298 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:59,293 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:28:00,940 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.763e+02 3.478e+02 4.434e+02 7.343e+02, threshold=6.957e+02, percent-clipped=3.0 2022-12-07 12:28:26,685 INFO [train.py:873] (1/4) Epoch 5, batch 2400, loss[loss=0.1817, simple_loss=0.1609, pruned_loss=0.1013, over 1291.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.1941, pruned_loss=0.08612, over 1976267.45 frames. ], batch size: 100, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:28:30,926 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:29:31,401 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.127e+02 2.520e+02 3.446e+02 4.436e+02 8.213e+02, threshold=6.892e+02, percent-clipped=3.0 2022-12-07 12:29:49,019 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1157, 1.9852, 1.7262, 1.7823, 2.0078, 1.9153, 2.0798, 2.0045], device='cuda:1'), covar=tensor([0.0893, 0.1249, 0.2461, 0.3031, 0.1008, 0.1151, 0.1477, 0.1145], device='cuda:1'), in_proj_covar=tensor([0.0268, 0.0233, 0.0345, 0.0436, 0.0261, 0.0304, 0.0321, 0.0266], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:29:56,007 INFO [train.py:873] (1/4) Epoch 5, batch 2500, loss[loss=0.1717, simple_loss=0.1844, pruned_loss=0.07949, over 6009.00 frames. ], tot_loss[loss=0.181, simple_loss=0.1931, pruned_loss=0.08446, over 1992026.53 frames. ], batch size: 100, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:30:57,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8716, 1.5134, 3.8351, 1.6317, 3.9342, 3.9980, 3.0326, 4.2957], device='cuda:1'), covar=tensor([0.0166, 0.2545, 0.0282, 0.2031, 0.0245, 0.0249, 0.0456, 0.0127], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0155, 0.0131, 0.0165, 0.0146, 0.0139, 0.0116, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:31:01,909 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.569e+02 3.218e+02 4.442e+02 9.627e+02, threshold=6.437e+02, percent-clipped=1.0 2022-12-07 12:31:27,904 INFO [train.py:873] (1/4) Epoch 5, batch 2600, loss[loss=0.2135, simple_loss=0.2124, pruned_loss=0.1073, over 7793.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.1929, pruned_loss=0.08486, over 1958540.26 frames. ], batch size: 100, lr: 1.57e-02, grad_scale: 16.0 2022-12-07 12:31:35,967 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 12:32:17,950 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5126, 1.4696, 3.4714, 1.5614, 3.5680, 3.5015, 2.5017, 3.8818], device='cuda:1'), covar=tensor([0.0181, 0.2567, 0.0386, 0.2048, 0.0366, 0.0290, 0.0611, 0.0122], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0154, 0.0132, 0.0166, 0.0145, 0.0139, 0.0116, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:32:33,210 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.535e+02 3.293e+02 4.332e+02 8.259e+02, threshold=6.586e+02, percent-clipped=6.0 2022-12-07 12:32:57,303 INFO [train.py:873] (1/4) Epoch 5, batch 2700, loss[loss=0.1659, simple_loss=0.1657, pruned_loss=0.08304, over 3849.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.1923, pruned_loss=0.08396, over 2013748.30 frames. ], batch size: 100, lr: 1.57e-02, grad_scale: 8.0 2022-12-07 12:33:13,521 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:33:51,275 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.23 vs. limit=2.0 2022-12-07 12:33:52,614 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7646, 1.5975, 3.6373, 1.5680, 3.7496, 3.6479, 2.9463, 4.0518], device='cuda:1'), covar=tensor([0.0190, 0.2696, 0.0364, 0.2320, 0.0344, 0.0297, 0.0553, 0.0146], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0132, 0.0168, 0.0147, 0.0141, 0.0118, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:34:03,193 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.537e+02 3.172e+02 3.779e+02 6.676e+02, threshold=6.344e+02, percent-clipped=1.0 2022-12-07 12:34:07,330 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:34:14,346 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9853, 4.5249, 4.3964, 4.9464, 4.5813, 4.3053, 4.9000, 4.1468], device='cuda:1'), covar=tensor([0.0314, 0.0916, 0.0287, 0.0371, 0.0719, 0.0488, 0.0478, 0.0433], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0212, 0.0143, 0.0136, 0.0140, 0.0114, 0.0212, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 12:34:27,437 INFO [train.py:873] (1/4) Epoch 5, batch 2800, loss[loss=0.1828, simple_loss=0.2037, pruned_loss=0.08098, over 14065.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.1934, pruned_loss=0.08441, over 1992060.41 frames. ], batch size: 29, lr: 1.57e-02, grad_scale: 8.0 2022-12-07 12:34:45,049 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:34:55,813 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2736, 0.9260, 1.2029, 1.0420, 1.0802, 0.8024, 0.9496, 1.3803], device='cuda:1'), covar=tensor([0.0751, 0.0824, 0.0724, 0.1167, 0.0859, 0.0355, 0.0734, 0.1219], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0019, 0.0018, 0.0019, 0.0025, 0.0019, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.1738e-05, 7.4562e-05, 6.8529e-05, 7.1293e-05, 7.4002e-05, 9.1446e-05, 7.7198e-05, 6.9903e-05], device='cuda:1') 2022-12-07 12:35:24,626 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 12:35:33,622 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.428e+02 3.052e+02 4.047e+02 7.917e+02, threshold=6.105e+02, percent-clipped=1.0 2022-12-07 12:35:40,392 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:35:56,746 INFO [train.py:873] (1/4) Epoch 5, batch 2900, loss[loss=0.1781, simple_loss=0.1956, pruned_loss=0.08033, over 14575.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.1932, pruned_loss=0.08392, over 2006650.61 frames. ], batch size: 22, lr: 1.57e-02, grad_scale: 4.0 2022-12-07 12:35:57,339 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 12:36:01,983 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:36:19,850 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 12:36:26,652 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 12:36:51,541 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2013, 3.3080, 4.0868, 2.8628, 2.7455, 3.1271, 2.0686, 3.1283], device='cuda:1'), covar=tensor([0.1451, 0.0730, 0.0674, 0.3018, 0.2454, 0.1224, 0.5362, 0.1453], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0079, 0.0076, 0.0087, 0.0111, 0.0067, 0.0144, 0.0074], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 12:36:56,516 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:37:02,204 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 2.752e+02 3.414e+02 4.319e+02 7.862e+02, threshold=6.827e+02, percent-clipped=8.0 2022-12-07 12:37:23,617 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6463, 1.7372, 2.4384, 2.1903, 2.4758, 2.2977, 2.1899, 2.0341], device='cuda:1'), covar=tensor([0.0349, 0.2256, 0.0458, 0.1378, 0.0412, 0.0742, 0.0630, 0.1553], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0343, 0.0341, 0.0322, 0.0347, 0.0273, 0.0321, 0.0360], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:37:26,083 INFO [train.py:873] (1/4) Epoch 5, batch 3000, loss[loss=0.161, simple_loss=0.1827, pruned_loss=0.06965, over 13957.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.1936, pruned_loss=0.08507, over 1964509.39 frames. ], batch size: 26, lr: 1.56e-02, grad_scale: 4.0 2022-12-07 12:37:26,083 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 12:37:38,397 INFO [train.py:905] (1/4) Epoch 5, validation: loss=0.1238, simple_loss=0.167, pruned_loss=0.04033, over 857387.00 frames. 2022-12-07 12:37:38,399 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 12:37:53,958 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 12:38:06,927 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2022-12-07 12:38:31,532 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-07 12:38:44,248 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:38:44,934 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.483e+02 3.391e+02 4.332e+02 8.491e+02, threshold=6.783e+02, percent-clipped=1.0 2022-12-07 12:39:08,696 INFO [train.py:873] (1/4) Epoch 5, batch 3100, loss[loss=0.2081, simple_loss=0.1784, pruned_loss=0.1189, over 1197.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.1935, pruned_loss=0.08419, over 1990543.64 frames. ], batch size: 100, lr: 1.56e-02, grad_scale: 4.0 2022-12-07 12:39:54,799 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3171, 2.6527, 3.7908, 3.0986, 3.6842, 3.6839, 3.6957, 3.1390], device='cuda:1'), covar=tensor([0.0296, 0.2795, 0.0559, 0.1532, 0.0753, 0.0668, 0.1277, 0.1668], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0342, 0.0345, 0.0318, 0.0350, 0.0275, 0.0319, 0.0358], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:40:15,024 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.236e+02 3.030e+02 3.906e+02 6.925e+02, threshold=6.060e+02, percent-clipped=1.0 2022-12-07 12:40:17,248 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:40:38,099 INFO [train.py:873] (1/4) Epoch 5, batch 3200, loss[loss=0.1885, simple_loss=0.2072, pruned_loss=0.08488, over 14279.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.1937, pruned_loss=0.08478, over 2007242.05 frames. ], batch size: 66, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:40:50,239 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7022, 1.6698, 2.8328, 2.0820, 2.5728, 1.6723, 2.1986, 2.5760], device='cuda:1'), covar=tensor([0.0748, 0.4446, 0.0361, 0.5413, 0.0437, 0.3604, 0.1149, 0.0312], device='cuda:1'), in_proj_covar=tensor([0.0226, 0.0268, 0.0169, 0.0362, 0.0175, 0.0272, 0.0247, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:40:59,518 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:41:31,910 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9057, 4.8164, 4.5225, 5.0271, 4.5298, 4.1657, 4.9539, 4.8812], device='cuda:1'), covar=tensor([0.0648, 0.0658, 0.0578, 0.0477, 0.0602, 0.0451, 0.0649, 0.0656], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0093, 0.0107, 0.0105, 0.0112, 0.0085, 0.0123, 0.0105], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:41:34,528 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:41:44,536 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.667e+02 3.351e+02 4.253e+02 1.055e+03, threshold=6.703e+02, percent-clipped=3.0 2022-12-07 12:41:53,260 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:41:58,447 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8927, 1.6457, 2.0757, 1.6871, 2.0538, 1.4466, 1.6934, 1.7623], device='cuda:1'), covar=tensor([0.0695, 0.1519, 0.0122, 0.0956, 0.0311, 0.0862, 0.0507, 0.0206], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0269, 0.0167, 0.0353, 0.0172, 0.0267, 0.0244, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:42:05,547 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7394, 1.9191, 2.6663, 2.8251, 2.6547, 1.9676, 2.8176, 2.2287], device='cuda:1'), covar=tensor([0.0094, 0.0219, 0.0167, 0.0093, 0.0094, 0.0362, 0.0068, 0.0256], device='cuda:1'), in_proj_covar=tensor([0.0183, 0.0193, 0.0278, 0.0223, 0.0178, 0.0239, 0.0163, 0.0230], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:42:08,400 INFO [train.py:873] (1/4) Epoch 5, batch 3300, loss[loss=0.1461, simple_loss=0.1476, pruned_loss=0.07231, over 2578.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.193, pruned_loss=0.08482, over 1946898.51 frames. ], batch size: 100, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:43:13,014 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:43:13,762 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.043e+02 2.484e+02 3.236e+02 4.553e+02 1.226e+03, threshold=6.473e+02, percent-clipped=4.0 2022-12-07 12:43:37,538 INFO [train.py:873] (1/4) Epoch 5, batch 3400, loss[loss=0.141, simple_loss=0.1624, pruned_loss=0.05977, over 14576.00 frames. ], tot_loss[loss=0.181, simple_loss=0.1927, pruned_loss=0.08468, over 1927350.01 frames. ], batch size: 21, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:43:52,844 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7145, 3.5635, 3.9086, 3.4169, 3.6741, 3.8225, 1.3436, 3.5687], device='cuda:1'), covar=tensor([0.0194, 0.0304, 0.0344, 0.0368, 0.0271, 0.0261, 0.3196, 0.0221], device='cuda:1'), in_proj_covar=tensor([0.0124, 0.0135, 0.0127, 0.0108, 0.0165, 0.0114, 0.0151, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:43:52,907 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3271, 2.3769, 2.2212, 2.4669, 1.8056, 2.3515, 2.1663, 1.0790], device='cuda:1'), covar=tensor([0.2133, 0.0933, 0.0859, 0.0575, 0.1165, 0.0818, 0.1277, 0.4182], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0061, 0.0052, 0.0051, 0.0075, 0.0054, 0.0084, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:43:55,870 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0045, 0.8889, 0.9738, 0.9906, 0.8871, 0.7814, 1.0471, 0.7422], device='cuda:1'), covar=tensor([0.0465, 0.0711, 0.0504, 0.0389, 0.0456, 0.0499, 0.0221, 0.0720], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0011, 0.0012, 0.0016, 0.0011, 0.0017], device='cuda:1'), out_proj_covar=tensor([5.7794e-05, 5.9277e-05, 5.9127e-05, 5.4637e-05, 6.1393e-05, 8.1918e-05, 6.3896e-05, 8.0420e-05], device='cuda:1') 2022-12-07 12:43:57,508 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:03,512 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:31,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5804, 2.4106, 2.4238, 1.5468, 2.3270, 2.6254, 2.7951, 2.2531], device='cuda:1'), covar=tensor([0.0641, 0.1698, 0.1049, 0.2905, 0.1005, 0.0463, 0.0494, 0.1603], device='cuda:1'), in_proj_covar=tensor([0.0102, 0.0215, 0.0119, 0.0133, 0.0102, 0.0105, 0.0089, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:44:37,024 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4181, 2.1892, 2.9063, 2.0204, 1.9148, 2.4966, 1.3069, 2.3539], device='cuda:1'), covar=tensor([0.0796, 0.1153, 0.0519, 0.1557, 0.1938, 0.1042, 0.4793, 0.0859], device='cuda:1'), in_proj_covar=tensor([0.0069, 0.0075, 0.0073, 0.0082, 0.0101, 0.0063, 0.0134, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 12:44:43,755 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.478e+02 3.354e+02 4.337e+02 1.168e+03, threshold=6.709e+02, percent-clipped=6.0 2022-12-07 12:44:45,715 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:50,837 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7494, 2.5844, 2.7623, 2.7673, 2.6984, 2.4854, 1.4463, 2.4582], device='cuda:1'), covar=tensor([0.0279, 0.0343, 0.0428, 0.0249, 0.0285, 0.0732, 0.2401, 0.0324], device='cuda:1'), in_proj_covar=tensor([0.0124, 0.0134, 0.0127, 0.0108, 0.0164, 0.0114, 0.0151, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:44:56,997 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:45:07,676 INFO [train.py:873] (1/4) Epoch 5, batch 3500, loss[loss=0.153, simple_loss=0.181, pruned_loss=0.06249, over 14268.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.1925, pruned_loss=0.08437, over 2016235.37 frames. ], batch size: 57, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:45:12,342 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1712, 2.0490, 3.0155, 2.3510, 2.9148, 2.8004, 2.7771, 2.4185], device='cuda:1'), covar=tensor([0.0365, 0.2916, 0.0728, 0.1734, 0.0514, 0.0766, 0.0872, 0.1928], device='cuda:1'), in_proj_covar=tensor([0.0265, 0.0342, 0.0346, 0.0317, 0.0343, 0.0275, 0.0323, 0.0360], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:45:22,382 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0385, 2.1194, 2.0098, 2.1713, 1.7563, 2.2447, 1.8891, 1.0228], device='cuda:1'), covar=tensor([0.2607, 0.1025, 0.1132, 0.0717, 0.1284, 0.0584, 0.1946, 0.4115], device='cuda:1'), in_proj_covar=tensor([0.0161, 0.0061, 0.0053, 0.0052, 0.0076, 0.0055, 0.0085, 0.0101], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:45:28,348 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:02,459 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:03,434 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5276, 2.1385, 3.5110, 3.6257, 3.5957, 2.2825, 3.5644, 2.5003], device='cuda:1'), covar=tensor([0.0122, 0.0363, 0.0340, 0.0137, 0.0118, 0.0528, 0.0088, 0.0460], device='cuda:1'), in_proj_covar=tensor([0.0182, 0.0191, 0.0276, 0.0222, 0.0177, 0.0237, 0.0163, 0.0230], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:46:13,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.895e+01 2.306e+02 2.839e+02 3.644e+02 6.341e+02, threshold=5.678e+02, percent-clipped=0.0 2022-12-07 12:46:13,578 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-07 12:46:18,218 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:32,574 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6471, 1.8405, 2.4563, 2.2177, 2.5504, 2.4505, 2.1908, 2.1881], device='cuda:1'), covar=tensor([0.0250, 0.1795, 0.0485, 0.1019, 0.0298, 0.0463, 0.0425, 0.1074], device='cuda:1'), in_proj_covar=tensor([0.0263, 0.0334, 0.0341, 0.0311, 0.0338, 0.0271, 0.0319, 0.0351], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 12:46:36,551 INFO [train.py:873] (1/4) Epoch 5, batch 3600, loss[loss=0.1896, simple_loss=0.1967, pruned_loss=0.09121, over 14248.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.1917, pruned_loss=0.08335, over 1985073.00 frames. ], batch size: 69, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:46:45,903 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:56,244 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0979, 4.8593, 4.7930, 5.1344, 4.7509, 4.2624, 5.2222, 5.0113], device='cuda:1'), covar=tensor([0.0626, 0.0490, 0.0439, 0.0450, 0.0523, 0.0384, 0.0513, 0.0543], device='cuda:1'), in_proj_covar=tensor([0.0110, 0.0094, 0.0109, 0.0108, 0.0116, 0.0087, 0.0124, 0.0107], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:47:32,634 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3090, 1.2597, 1.2302, 1.2806, 1.1164, 0.8610, 1.4729, 0.8789], device='cuda:1'), covar=tensor([0.0691, 0.0788, 0.0928, 0.1737, 0.0812, 0.0299, 0.0293, 0.0895], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0011, 0.0011, 0.0010, 0.0011, 0.0015, 0.0011, 0.0016], device='cuda:1'), out_proj_covar=tensor([5.5939e-05, 5.6647e-05, 5.6541e-05, 5.1756e-05, 5.7739e-05, 7.5506e-05, 6.0360e-05, 7.6428e-05], device='cuda:1') 2022-12-07 12:47:39,033 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7970, 1.4285, 2.0836, 1.3096, 2.0709, 2.0960, 1.6790, 2.0940], device='cuda:1'), covar=tensor([0.0234, 0.1175, 0.0209, 0.1063, 0.0230, 0.0245, 0.0587, 0.0192], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0162, 0.0133, 0.0170, 0.0152, 0.0148, 0.0123, 0.0122], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:47:44,853 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.078e+02 2.977e+02 3.494e+02 4.554e+02 8.959e+02, threshold=6.988e+02, percent-clipped=9.0 2022-12-07 12:47:59,706 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:48:09,172 INFO [train.py:873] (1/4) Epoch 5, batch 3700, loss[loss=0.2152, simple_loss=0.2144, pruned_loss=0.108, over 14085.00 frames. ], tot_loss[loss=0.181, simple_loss=0.1928, pruned_loss=0.0846, over 1948967.91 frames. ], batch size: 26, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:48:19,529 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9855, 2.8847, 2.9985, 2.8660, 2.9226, 2.7188, 1.2790, 2.7098], device='cuda:1'), covar=tensor([0.0235, 0.0304, 0.0400, 0.0375, 0.0286, 0.0566, 0.2778, 0.0276], device='cuda:1'), in_proj_covar=tensor([0.0123, 0.0133, 0.0126, 0.0109, 0.0165, 0.0116, 0.0151, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:48:31,903 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-07 12:48:54,447 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:49:15,044 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.468e+02 3.019e+02 3.918e+02 8.852e+02, threshold=6.038e+02, percent-clipped=4.0 2022-12-07 12:49:24,197 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:49:34,013 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:49:38,093 INFO [train.py:873] (1/4) Epoch 5, batch 3800, loss[loss=0.1619, simple_loss=0.1831, pruned_loss=0.07029, over 14284.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.1916, pruned_loss=0.08381, over 1966480.95 frames. ], batch size: 63, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:49:43,528 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9204, 1.5414, 3.3955, 3.2236, 3.3010, 3.4705, 2.8815, 3.4223], device='cuda:1'), covar=tensor([0.1051, 0.1175, 0.0092, 0.0174, 0.0166, 0.0088, 0.0185, 0.0106], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0160, 0.0102, 0.0142, 0.0119, 0.0122, 0.0091, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:50:27,315 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7456, 2.1728, 3.4331, 2.4866, 3.4279, 3.4467, 3.1958, 2.6582], device='cuda:1'), covar=tensor([0.0407, 0.3216, 0.0694, 0.2036, 0.0769, 0.0592, 0.1808, 0.3022], device='cuda:1'), in_proj_covar=tensor([0.0265, 0.0338, 0.0348, 0.0314, 0.0343, 0.0278, 0.0325, 0.0362], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:50:29,162 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:50:45,933 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.542e+02 3.059e+02 3.773e+02 1.009e+03, threshold=6.118e+02, percent-clipped=4.0 2022-12-07 12:50:50,686 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:50:51,810 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-07 12:51:05,962 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:51:09,662 INFO [train.py:873] (1/4) Epoch 5, batch 3900, loss[loss=0.1693, simple_loss=0.1953, pruned_loss=0.07168, over 14601.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.1921, pruned_loss=0.08332, over 2026729.49 frames. ], batch size: 22, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:51:33,817 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:51:50,721 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.50 vs. limit=2.0 2022-12-07 12:52:02,414 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:52:17,180 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0203, 2.6926, 5.0293, 3.3558, 4.9546, 2.2900, 3.8346, 4.5396], device='cuda:1'), covar=tensor([0.0289, 0.4636, 0.0309, 0.8930, 0.0179, 0.4023, 0.1051, 0.0187], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0262, 0.0167, 0.0351, 0.0173, 0.0263, 0.0247, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:52:17,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.452e+02 3.217e+02 4.044e+02 1.041e+03, threshold=6.433e+02, percent-clipped=5.0 2022-12-07 12:52:31,224 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 12:52:35,160 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6707, 5.4182, 5.1496, 5.7259, 5.1975, 5.0294, 5.7051, 5.6357], device='cuda:1'), covar=tensor([0.0608, 0.0407, 0.0570, 0.0413, 0.0534, 0.0379, 0.0655, 0.0623], device='cuda:1'), in_proj_covar=tensor([0.0107, 0.0094, 0.0109, 0.0108, 0.0115, 0.0088, 0.0124, 0.0108], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:52:36,437 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-07 12:52:41,186 INFO [train.py:873] (1/4) Epoch 5, batch 4000, loss[loss=0.1747, simple_loss=0.1824, pruned_loss=0.08354, over 6006.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.1912, pruned_loss=0.08197, over 2078624.41 frames. ], batch size: 100, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:53:07,704 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3417, 2.1586, 3.2751, 1.9281, 2.2918, 2.5277, 1.1378, 2.4041], device='cuda:1'), covar=tensor([0.1362, 0.1066, 0.0591, 0.2576, 0.2119, 0.0971, 0.5771, 0.1064], device='cuda:1'), in_proj_covar=tensor([0.0069, 0.0076, 0.0073, 0.0081, 0.0102, 0.0064, 0.0137, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 12:53:22,798 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:53:25,711 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6783, 2.5238, 4.6522, 3.2547, 4.5129, 2.1709, 3.2498, 4.2641], device='cuda:1'), covar=tensor([0.0302, 0.4645, 0.0527, 0.8379, 0.0261, 0.3886, 0.1370, 0.0202], device='cuda:1'), in_proj_covar=tensor([0.0222, 0.0266, 0.0166, 0.0350, 0.0175, 0.0266, 0.0252, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:53:48,675 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.578e+02 3.207e+02 3.995e+02 8.945e+02, threshold=6.415e+02, percent-clipped=4.0 2022-12-07 12:53:57,371 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:54:03,098 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9551, 5.3542, 5.4723, 5.9300, 5.6383, 4.8992, 5.9548, 4.8035], device='cuda:1'), covar=tensor([0.0234, 0.0836, 0.0235, 0.0338, 0.0558, 0.0277, 0.0358, 0.0405], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0216, 0.0142, 0.0133, 0.0138, 0.0111, 0.0210, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 12:54:12,175 INFO [train.py:873] (1/4) Epoch 5, batch 4100, loss[loss=0.2053, simple_loss=0.1875, pruned_loss=0.1115, over 2611.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.1912, pruned_loss=0.08265, over 1954325.36 frames. ], batch size: 100, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:54:34,318 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0574, 0.7427, 1.0440, 0.8090, 1.1054, 0.8232, 1.0588, 1.0151], device='cuda:1'), covar=tensor([0.1175, 0.1086, 0.0595, 0.0902, 0.1258, 0.0616, 0.0810, 0.1345], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0018, 0.0018, 0.0020, 0.0026, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([7.1869e-05, 7.5367e-05, 6.8376e-05, 7.2077e-05, 7.6560e-05, 9.5440e-05, 8.0127e-05, 7.3856e-05], device='cuda:1') 2022-12-07 12:54:40,229 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:54:57,937 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:55:18,131 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.334e+02 3.145e+02 3.965e+02 8.915e+02, threshold=6.291e+02, percent-clipped=4.0 2022-12-07 12:55:36,587 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1424, 1.2831, 1.5085, 0.9872, 0.9617, 1.2341, 0.6727, 1.1601], device='cuda:1'), covar=tensor([0.1562, 0.2700, 0.0627, 0.2173, 0.3282, 0.0551, 0.1849, 0.0976], device='cuda:1'), in_proj_covar=tensor([0.0070, 0.0078, 0.0074, 0.0083, 0.0106, 0.0065, 0.0138, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 12:55:41,760 INFO [train.py:873] (1/4) Epoch 5, batch 4200, loss[loss=0.1719, simple_loss=0.1846, pruned_loss=0.07961, over 12017.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.192, pruned_loss=0.08368, over 1961322.37 frames. ], batch size: 100, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:56:24,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-07 12:56:28,631 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:56:34,768 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.83 vs. limit=5.0 2022-12-07 12:56:48,745 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.089e+02 2.566e+02 3.165e+02 3.922e+02 1.587e+03, threshold=6.330e+02, percent-clipped=5.0 2022-12-07 12:57:00,648 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9548, 1.5341, 3.9466, 3.7001, 3.8112, 3.9842, 3.3972, 4.0207], device='cuda:1'), covar=tensor([0.0973, 0.1122, 0.0073, 0.0122, 0.0119, 0.0070, 0.0144, 0.0076], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0155, 0.0100, 0.0139, 0.0117, 0.0121, 0.0091, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 12:57:09,320 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0200, 3.4816, 2.6099, 4.1248, 3.9664, 3.9263, 3.2294, 2.6423], device='cuda:1'), covar=tensor([0.0648, 0.1362, 0.4244, 0.0339, 0.0594, 0.0696, 0.1239, 0.4596], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0303, 0.0300, 0.0185, 0.0245, 0.0245, 0.0258, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:57:11,679 INFO [train.py:873] (1/4) Epoch 5, batch 4300, loss[loss=0.1505, simple_loss=0.1848, pruned_loss=0.05804, over 14002.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.1922, pruned_loss=0.08377, over 1959260.78 frames. ], batch size: 26, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:57:51,397 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:57:53,630 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:17,777 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:18,607 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.121e+02 2.611e+02 3.299e+02 4.173e+02 8.760e+02, threshold=6.598e+02, percent-clipped=4.0 2022-12-07 12:58:18,871 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7494, 2.0369, 2.8377, 2.8229, 2.7134, 1.9406, 3.0194, 2.2624], device='cuda:1'), covar=tensor([0.0107, 0.0315, 0.0246, 0.0180, 0.0120, 0.0492, 0.0076, 0.0380], device='cuda:1'), in_proj_covar=tensor([0.0189, 0.0192, 0.0281, 0.0228, 0.0178, 0.0239, 0.0166, 0.0232], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 12:58:21,424 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:36,755 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:41,837 INFO [train.py:873] (1/4) Epoch 5, batch 4400, loss[loss=0.1994, simple_loss=0.178, pruned_loss=0.1104, over 2627.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.1925, pruned_loss=0.08397, over 1969007.55 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 12:58:46,274 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:12,448 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:59:16,032 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:20,632 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 12:59:27,209 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:32,315 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8968, 2.0001, 4.1191, 2.8203, 3.9502, 1.8833, 2.8748, 3.7768], device='cuda:1'), covar=tensor([0.0429, 0.5196, 0.0235, 0.8259, 0.0284, 0.4030, 0.1481, 0.0258], device='cuda:1'), in_proj_covar=tensor([0.0219, 0.0255, 0.0164, 0.0346, 0.0173, 0.0263, 0.0245, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 12:59:35,661 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 12:59:36,156 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:59:47,496 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2168, 1.6386, 1.5559, 1.6960, 1.4635, 1.7137, 1.3686, 0.9729], device='cuda:1'), covar=tensor([0.2327, 0.0704, 0.0564, 0.0439, 0.0757, 0.0552, 0.1544, 0.2461], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0057, 0.0052, 0.0049, 0.0071, 0.0055, 0.0081, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 12:59:47,543 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6934, 2.4994, 1.7997, 2.7421, 2.4837, 2.5545, 2.3036, 1.9593], device='cuda:1'), covar=tensor([0.0526, 0.0919, 0.2952, 0.0252, 0.0881, 0.0894, 0.1202, 0.2982], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0302, 0.0297, 0.0186, 0.0245, 0.0245, 0.0256, 0.0292], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 12:59:48,178 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.387e+02 3.073e+02 3.868e+02 1.059e+03, threshold=6.147e+02, percent-clipped=1.0 2022-12-07 13:00:10,389 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:00:11,140 INFO [train.py:873] (1/4) Epoch 5, batch 4500, loss[loss=0.2224, simple_loss=0.19, pruned_loss=0.1274, over 1282.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.1915, pruned_loss=0.0831, over 1971165.00 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:00:31,245 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:00:37,552 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 13:00:38,327 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 13:00:57,949 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:01:16,934 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.392e+02 3.325e+02 4.104e+02 6.523e+02, threshold=6.650e+02, percent-clipped=2.0 2022-12-07 13:01:20,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9538, 3.5133, 2.6143, 4.0952, 3.9187, 3.9040, 3.2261, 2.8200], device='cuda:1'), covar=tensor([0.0519, 0.1086, 0.3690, 0.0242, 0.0561, 0.1046, 0.1097, 0.3496], device='cuda:1'), in_proj_covar=tensor([0.0228, 0.0306, 0.0299, 0.0189, 0.0249, 0.0250, 0.0261, 0.0292], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:01:37,268 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4483, 1.8415, 3.6043, 2.3994, 3.4248, 1.8418, 2.6944, 3.1975], device='cuda:1'), covar=tensor([0.0734, 0.5285, 0.0492, 0.7717, 0.0405, 0.4485, 0.1389, 0.0504], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0259, 0.0166, 0.0349, 0.0174, 0.0265, 0.0242, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:01:40,483 INFO [train.py:873] (1/4) Epoch 5, batch 4600, loss[loss=0.2019, simple_loss=0.2009, pruned_loss=0.1015, over 7803.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.1911, pruned_loss=0.08231, over 1998404.51 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:01:40,564 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:01:41,008 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.63 vs. limit=2.0 2022-12-07 13:01:51,946 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 13:02:29,001 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 13:02:46,773 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 2.771e+02 3.403e+02 4.151e+02 6.915e+02, threshold=6.807e+02, percent-clipped=1.0 2022-12-07 13:02:47,435 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 13:03:10,234 INFO [train.py:873] (1/4) Epoch 5, batch 4700, loss[loss=0.1747, simple_loss=0.1691, pruned_loss=0.09016, over 3813.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.191, pruned_loss=0.08223, over 1986237.97 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:03:10,351 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:03:32,449 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4566, 2.3591, 4.4732, 3.0033, 4.2961, 2.1387, 3.2511, 4.2020], device='cuda:1'), covar=tensor([0.0324, 0.4702, 0.0294, 0.8844, 0.0332, 0.3850, 0.1050, 0.0237], device='cuda:1'), in_proj_covar=tensor([0.0219, 0.0260, 0.0164, 0.0350, 0.0173, 0.0264, 0.0239, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:03:35,668 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:03:39,283 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:04:03,465 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0819, 2.5065, 5.0698, 3.3075, 4.9379, 2.5259, 3.7666, 4.6360], device='cuda:1'), covar=tensor([0.0313, 0.5272, 0.0526, 1.1196, 0.0195, 0.3809, 0.1151, 0.0222], device='cuda:1'), in_proj_covar=tensor([0.0220, 0.0262, 0.0166, 0.0353, 0.0173, 0.0266, 0.0239, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:04:21,150 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.441e+02 3.432e+02 4.388e+02 1.189e+03, threshold=6.863e+02, percent-clipped=8.0 2022-12-07 13:04:45,219 INFO [train.py:873] (1/4) Epoch 5, batch 4800, loss[loss=0.1424, simple_loss=0.1723, pruned_loss=0.05632, over 14643.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.1907, pruned_loss=0.08207, over 1985785.88 frames. ], batch size: 23, lr: 1.52e-02, grad_scale: 8.0 2022-12-07 13:05:01,117 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:05:02,036 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9017, 2.6000, 2.6622, 2.8468, 2.7585, 2.7989, 2.9092, 2.3727], device='cuda:1'), covar=tensor([0.0384, 0.1009, 0.0460, 0.0504, 0.0633, 0.0392, 0.0669, 0.0605], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0218, 0.0144, 0.0136, 0.0140, 0.0111, 0.0210, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 13:05:24,555 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:05:24,620 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3779, 2.9316, 4.1553, 3.0418, 4.0924, 4.0301, 3.8893, 3.5582], device='cuda:1'), covar=tensor([0.0254, 0.2266, 0.0677, 0.1515, 0.0598, 0.0532, 0.1644, 0.1547], device='cuda:1'), in_proj_covar=tensor([0.0269, 0.0333, 0.0350, 0.0308, 0.0336, 0.0276, 0.0321, 0.0350], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 13:05:38,673 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7008, 3.0326, 4.5252, 3.1799, 4.3606, 4.3508, 4.1790, 3.8737], device='cuda:1'), covar=tensor([0.0228, 0.2475, 0.0608, 0.1724, 0.0569, 0.0534, 0.1615, 0.1904], device='cuda:1'), in_proj_covar=tensor([0.0270, 0.0335, 0.0352, 0.0311, 0.0340, 0.0278, 0.0322, 0.0353], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 13:05:53,123 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.350e+02 3.192e+02 4.148e+02 1.074e+03, threshold=6.385e+02, percent-clipped=3.0 2022-12-07 13:06:03,854 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 13:06:16,622 INFO [train.py:873] (1/4) Epoch 5, batch 4900, loss[loss=0.1684, simple_loss=0.1987, pruned_loss=0.06903, over 14555.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.1919, pruned_loss=0.083, over 1984536.19 frames. ], batch size: 43, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:06:20,593 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:06:33,127 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1148, 2.0610, 1.8007, 1.7716, 2.0820, 1.9982, 2.0594, 2.0148], device='cuda:1'), covar=tensor([0.0847, 0.0801, 0.1994, 0.2656, 0.0814, 0.0866, 0.1293, 0.1099], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0222, 0.0348, 0.0432, 0.0254, 0.0310, 0.0327, 0.0269], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:07:23,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8391, 0.7323, 0.9735, 1.4587, 0.8989, 0.6486, 1.2487, 1.3338], device='cuda:1'), covar=tensor([0.2139, 0.1184, 0.0683, 0.1774, 0.1855, 0.0553, 0.0801, 0.1702], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0018, 0.0018, 0.0019, 0.0024, 0.0019, 0.0018], device='cuda:1'), out_proj_covar=tensor([6.9993e-05, 7.2819e-05, 6.7268e-05, 7.0841e-05, 7.4975e-05, 9.0146e-05, 7.7692e-05, 7.2572e-05], device='cuda:1') 2022-12-07 13:07:24,093 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.440e+01 2.537e+02 3.134e+02 4.134e+02 8.105e+02, threshold=6.269e+02, percent-clipped=2.0 2022-12-07 13:07:47,530 INFO [train.py:873] (1/4) Epoch 5, batch 5000, loss[loss=0.1988, simple_loss=0.1963, pruned_loss=0.1006, over 7742.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.1904, pruned_loss=0.08166, over 1938371.20 frames. ], batch size: 100, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:07:47,686 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:08,694 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7912, 2.0725, 3.6904, 3.8358, 3.8116, 2.3449, 3.7258, 2.9514], device='cuda:1'), covar=tensor([0.0127, 0.0336, 0.0376, 0.0177, 0.0090, 0.0545, 0.0087, 0.0392], device='cuda:1'), in_proj_covar=tensor([0.0188, 0.0191, 0.0278, 0.0226, 0.0176, 0.0237, 0.0167, 0.0233], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 13:08:13,640 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:08:16,174 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4069, 2.1691, 4.4213, 2.9712, 4.2544, 2.1721, 3.1773, 4.1230], device='cuda:1'), covar=tensor([0.0307, 0.4819, 0.0196, 0.8923, 0.0236, 0.3696, 0.1212, 0.0173], device='cuda:1'), in_proj_covar=tensor([0.0220, 0.0263, 0.0166, 0.0354, 0.0175, 0.0268, 0.0243, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:08:17,371 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:30,434 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:53,563 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.768e+02 3.459e+02 4.682e+02 1.378e+03, threshold=6.919e+02, percent-clipped=12.0 2022-12-07 13:08:56,553 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:09:00,113 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:09:13,384 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:09:17,078 INFO [train.py:873] (1/4) Epoch 5, batch 5100, loss[loss=0.1472, simple_loss=0.178, pruned_loss=0.05814, over 14243.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.1902, pruned_loss=0.08175, over 1944185.53 frames. ], batch size: 35, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:09:32,870 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:09:39,900 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:09:47,490 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 13:10:08,801 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:10:12,330 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2022-12-07 13:10:15,185 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:10:23,148 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.613e+02 3.282e+02 4.367e+02 7.735e+02, threshold=6.564e+02, percent-clipped=1.0 2022-12-07 13:10:34,114 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:10:44,937 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:10:45,710 INFO [train.py:873] (1/4) Epoch 5, batch 5200, loss[loss=0.1608, simple_loss=0.1659, pruned_loss=0.07791, over 3848.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.1915, pruned_loss=0.08285, over 1969086.39 frames. ], batch size: 100, lr: 1.52e-02, grad_scale: 8.0 2022-12-07 13:11:51,443 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.356e+02 3.494e+02 4.184e+02 1.202e+03, threshold=6.988e+02, percent-clipped=4.0 2022-12-07 13:11:52,492 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1944, 1.5461, 1.6220, 1.5815, 1.5097, 1.6196, 1.2709, 0.9993], device='cuda:1'), covar=tensor([0.2257, 0.1702, 0.1117, 0.0434, 0.1194, 0.0715, 0.2006, 0.3914], device='cuda:1'), in_proj_covar=tensor([0.0159, 0.0062, 0.0053, 0.0052, 0.0072, 0.0057, 0.0084, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:11:57,340 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:12:13,730 INFO [train.py:873] (1/4) Epoch 5, batch 5300, loss[loss=0.2212, simple_loss=0.1783, pruned_loss=0.1321, over 1297.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1917, pruned_loss=0.08302, over 1961039.24 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:12:21,128 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8499, 2.7182, 3.5997, 2.3204, 2.4169, 2.7918, 1.6356, 2.7219], device='cuda:1'), covar=tensor([0.1065, 0.1555, 0.0617, 0.2311, 0.2348, 0.1065, 0.5121, 0.1225], device='cuda:1'), in_proj_covar=tensor([0.0067, 0.0078, 0.0072, 0.0080, 0.0103, 0.0064, 0.0131, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 13:12:51,570 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:12:56,305 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6923, 1.4706, 2.0415, 1.7059, 1.8965, 1.4560, 1.6130, 1.6468], device='cuda:1'), covar=tensor([0.0887, 0.2077, 0.0149, 0.0966, 0.0689, 0.1213, 0.0533, 0.0454], device='cuda:1'), in_proj_covar=tensor([0.0219, 0.0264, 0.0165, 0.0349, 0.0176, 0.0264, 0.0244, 0.0175], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:13:20,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.787e+02 3.586e+02 4.482e+02 8.195e+02, threshold=7.172e+02, percent-clipped=2.0 2022-12-07 13:13:31,248 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4139, 4.0194, 3.8728, 4.3974, 4.0736, 3.9326, 4.4121, 3.6493], device='cuda:1'), covar=tensor([0.0365, 0.1167, 0.0390, 0.0433, 0.0785, 0.0827, 0.0541, 0.0586], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0214, 0.0143, 0.0135, 0.0142, 0.0112, 0.0210, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 13:13:38,707 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3676, 3.7079, 2.9547, 4.7072, 4.1006, 4.4816, 3.4947, 3.1869], device='cuda:1'), covar=tensor([0.0521, 0.1114, 0.3697, 0.0263, 0.0861, 0.0976, 0.1241, 0.3233], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0297, 0.0293, 0.0184, 0.0247, 0.0249, 0.0254, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:13:43,972 INFO [train.py:873] (1/4) Epoch 5, batch 5400, loss[loss=0.1678, simple_loss=0.1923, pruned_loss=0.07165, over 14264.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.1917, pruned_loss=0.08236, over 1990779.32 frames. ], batch size: 57, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:14:30,637 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:14:51,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.811e+02 3.690e+02 4.403e+02 1.133e+03, threshold=7.380e+02, percent-clipped=4.0 2022-12-07 13:14:57,174 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:12,979 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:13,722 INFO [train.py:873] (1/4) Epoch 5, batch 5500, loss[loss=0.1866, simple_loss=0.1842, pruned_loss=0.09451, over 5973.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.1901, pruned_loss=0.08106, over 1963053.13 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:15:33,605 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:55,917 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:09,235 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:20,154 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.649e+02 3.424e+02 4.152e+02 8.919e+02, threshold=6.848e+02, percent-clipped=1.0 2022-12-07 13:16:23,442 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5165, 3.1221, 4.2486, 3.3358, 4.2394, 4.4016, 3.8685, 3.6746], device='cuda:1'), covar=tensor([0.0377, 0.2763, 0.0784, 0.1607, 0.0779, 0.0469, 0.2444, 0.2094], device='cuda:1'), in_proj_covar=tensor([0.0276, 0.0342, 0.0359, 0.0313, 0.0346, 0.0284, 0.0333, 0.0362], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:16:28,227 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:29,034 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1551, 1.4687, 4.5753, 4.1842, 4.2858, 4.6769, 4.4817, 4.6900], device='cuda:1'), covar=tensor([0.1177, 0.1365, 0.0090, 0.0131, 0.0115, 0.0088, 0.0069, 0.0089], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0157, 0.0104, 0.0144, 0.0117, 0.0120, 0.0093, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 13:16:42,803 INFO [train.py:873] (1/4) Epoch 5, batch 5600, loss[loss=0.2174, simple_loss=0.1831, pruned_loss=0.1258, over 1308.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.19, pruned_loss=0.0817, over 1904651.67 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:16:47,598 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2816, 2.9943, 4.1362, 3.0620, 4.0428, 4.1332, 3.8907, 3.5687], device='cuda:1'), covar=tensor([0.0280, 0.2563, 0.0727, 0.1632, 0.0769, 0.0458, 0.1513, 0.1725], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0339, 0.0357, 0.0311, 0.0344, 0.0281, 0.0329, 0.0356], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:16:51,005 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6635, 2.0076, 2.5762, 2.3409, 2.7271, 2.5232, 2.4645, 2.3535], device='cuda:1'), covar=tensor([0.0314, 0.2017, 0.0531, 0.1135, 0.0453, 0.0590, 0.0528, 0.1079], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0339, 0.0357, 0.0312, 0.0345, 0.0281, 0.0329, 0.0357], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:17:03,257 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:17:15,314 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:17:48,145 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.385e+02 3.015e+02 3.800e+02 5.921e+02, threshold=6.029e+02, percent-clipped=0.0 2022-12-07 13:18:10,021 INFO [train.py:873] (1/4) Epoch 5, batch 5700, loss[loss=0.1835, simple_loss=0.1743, pruned_loss=0.09632, over 3896.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1913, pruned_loss=0.08315, over 1931743.58 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:18:55,006 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 13:18:55,465 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:19:08,031 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.60 vs. limit=5.0 2022-12-07 13:19:15,101 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.574e+02 3.156e+02 4.047e+02 9.769e+02, threshold=6.311e+02, percent-clipped=9.0 2022-12-07 13:19:20,498 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:19:37,071 INFO [train.py:873] (1/4) Epoch 5, batch 5800, loss[loss=0.1804, simple_loss=0.1945, pruned_loss=0.08312, over 14232.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.1909, pruned_loss=0.08225, over 1940308.42 frames. ], batch size: 32, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:19:37,135 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:19:40,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9633, 3.6808, 3.2011, 2.3416, 3.2165, 3.5241, 3.8736, 2.9342], device='cuda:1'), covar=tensor([0.0390, 0.2308, 0.1024, 0.2141, 0.0572, 0.0398, 0.0804, 0.1321], device='cuda:1'), in_proj_covar=tensor([0.0107, 0.0210, 0.0119, 0.0126, 0.0101, 0.0108, 0.0089, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 13:20:02,747 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:20:41,975 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8852, 2.6151, 2.6521, 1.5447, 2.4356, 2.5773, 2.9397, 2.2615], device='cuda:1'), covar=tensor([0.0644, 0.1722, 0.1248, 0.2819, 0.0918, 0.0724, 0.0603, 0.1791], device='cuda:1'), in_proj_covar=tensor([0.0109, 0.0212, 0.0120, 0.0129, 0.0103, 0.0110, 0.0090, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 13:20:42,662 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.552e+02 3.407e+02 4.382e+02 1.034e+03, threshold=6.813e+02, percent-clipped=3.0 2022-12-07 13:20:45,482 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:20:52,203 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.04 vs. limit=5.0 2022-12-07 13:21:04,917 INFO [train.py:873] (1/4) Epoch 5, batch 5900, loss[loss=0.1795, simple_loss=0.1963, pruned_loss=0.08135, over 14265.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.19, pruned_loss=0.08144, over 1931902.18 frames. ], batch size: 76, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:21:20,643 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:32,818 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:37,053 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:41,743 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 13:22:10,496 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.756e+02 3.283e+02 4.209e+02 7.399e+02, threshold=6.567e+02, percent-clipped=1.0 2022-12-07 13:22:20,120 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:22:27,264 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:22:33,229 INFO [train.py:873] (1/4) Epoch 5, batch 6000, loss[loss=0.1794, simple_loss=0.1761, pruned_loss=0.09131, over 4991.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.19, pruned_loss=0.08177, over 1922048.71 frames. ], batch size: 100, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:22:33,229 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 13:22:45,290 INFO [train.py:905] (1/4) Epoch 5, validation: loss=0.1225, simple_loss=0.1661, pruned_loss=0.03949, over 857387.00 frames. 2022-12-07 13:22:45,291 INFO [train.py:906] (1/4) Maximum memory allocated so far is 17855MB 2022-12-07 13:23:51,459 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.659e+02 3.309e+02 3.881e+02 6.355e+02, threshold=6.617e+02, percent-clipped=0.0 2022-12-07 13:24:09,756 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0041, 2.8861, 2.0435, 3.0448, 2.7504, 2.8731, 2.5086, 2.2513], device='cuda:1'), covar=tensor([0.0377, 0.1068, 0.3329, 0.0247, 0.0681, 0.0769, 0.1311, 0.2909], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0298, 0.0292, 0.0184, 0.0248, 0.0253, 0.0250, 0.0285], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:24:13,083 INFO [train.py:873] (1/4) Epoch 5, batch 6100, loss[loss=0.2413, simple_loss=0.2244, pruned_loss=0.1291, over 9533.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.1905, pruned_loss=0.08228, over 1939168.74 frames. ], batch size: 100, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:24:19,308 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7670, 1.1957, 2.4883, 2.3385, 2.4321, 2.5146, 1.7498, 2.4998], device='cuda:1'), covar=tensor([0.0602, 0.0899, 0.0104, 0.0212, 0.0177, 0.0095, 0.0295, 0.0125], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0101, 0.0144, 0.0117, 0.0120, 0.0093, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 13:24:38,074 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 13:24:40,830 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0422, 1.9080, 2.0407, 2.0352, 1.9881, 1.8224, 1.2973, 1.7561], device='cuda:1'), covar=tensor([0.0361, 0.0370, 0.0420, 0.0218, 0.0278, 0.0657, 0.1525, 0.0353], device='cuda:1'), in_proj_covar=tensor([0.0130, 0.0135, 0.0125, 0.0111, 0.0168, 0.0117, 0.0149, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:24:41,665 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8900, 1.5157, 1.9991, 1.6839, 2.0998, 1.7981, 1.6786, 1.8657], device='cuda:1'), covar=tensor([0.0189, 0.0947, 0.0096, 0.0178, 0.0135, 0.0282, 0.0098, 0.0196], device='cuda:1'), in_proj_covar=tensor([0.0272, 0.0336, 0.0359, 0.0309, 0.0342, 0.0283, 0.0334, 0.0353], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:25:10,478 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9101, 1.5699, 4.0221, 3.7861, 3.8968, 4.1355, 3.4152, 4.0932], device='cuda:1'), covar=tensor([0.1170, 0.1257, 0.0087, 0.0137, 0.0117, 0.0080, 0.0208, 0.0096], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0154, 0.0100, 0.0142, 0.0115, 0.0119, 0.0092, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 13:25:16,062 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4681, 3.0458, 2.2494, 3.5912, 3.3943, 3.4078, 2.9708, 2.2976], device='cuda:1'), covar=tensor([0.0504, 0.1391, 0.3559, 0.0246, 0.0521, 0.0683, 0.1040, 0.4035], device='cuda:1'), in_proj_covar=tensor([0.0226, 0.0300, 0.0292, 0.0184, 0.0247, 0.0253, 0.0253, 0.0288], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:25:19,184 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.518e+02 3.418e+02 4.191e+02 1.416e+03, threshold=6.837e+02, percent-clipped=13.0 2022-12-07 13:25:21,968 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:25:40,682 INFO [train.py:873] (1/4) Epoch 5, batch 6200, loss[loss=0.1468, simple_loss=0.1717, pruned_loss=0.061, over 14248.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.1907, pruned_loss=0.08256, over 1902716.70 frames. ], batch size: 57, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:25:56,850 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:03,689 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:17,937 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2299, 1.9963, 3.5056, 2.4831, 3.4039, 1.8528, 2.7159, 3.2475], device='cuda:1'), covar=tensor([0.0480, 0.4690, 0.0357, 0.6738, 0.0365, 0.3881, 0.1270, 0.0354], device='cuda:1'), in_proj_covar=tensor([0.0218, 0.0259, 0.0166, 0.0341, 0.0176, 0.0264, 0.0240, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:26:38,813 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:46,426 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.596e+02 3.351e+02 4.378e+02 1.450e+03, threshold=6.701e+02, percent-clipped=3.0 2022-12-07 13:26:50,007 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 13:26:57,924 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:27:08,589 INFO [train.py:873] (1/4) Epoch 5, batch 6300, loss[loss=0.1519, simple_loss=0.1808, pruned_loss=0.06156, over 14453.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.1897, pruned_loss=0.08099, over 1973927.67 frames. ], batch size: 51, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:27:53,661 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6211, 2.3273, 3.7399, 3.7655, 3.8144, 2.2848, 3.7566, 2.7739], device='cuda:1'), covar=tensor([0.0137, 0.0397, 0.0323, 0.0167, 0.0111, 0.0630, 0.0092, 0.0513], device='cuda:1'), in_proj_covar=tensor([0.0195, 0.0196, 0.0285, 0.0230, 0.0183, 0.0242, 0.0177, 0.0237], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 13:27:57,540 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:00,260 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:03,976 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:14,423 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.701e+02 3.163e+02 3.734e+02 7.022e+02, threshold=6.325e+02, percent-clipped=2.0 2022-12-07 13:28:35,795 INFO [train.py:873] (1/4) Epoch 5, batch 6400, loss[loss=0.1871, simple_loss=0.2004, pruned_loss=0.08693, over 14178.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.1888, pruned_loss=0.08032, over 1989202.53 frames. ], batch size: 84, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:28:51,082 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:53,714 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:57,118 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:29:07,823 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2022-12-07 13:29:14,394 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8477, 2.8809, 3.7134, 2.6337, 2.3320, 3.1259, 1.5076, 2.9254], device='cuda:1'), covar=tensor([0.1182, 0.1004, 0.0622, 0.2040, 0.2204, 0.1096, 0.5517, 0.1273], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0078, 0.0076, 0.0086, 0.0106, 0.0068, 0.0137, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:29:20,786 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:29:41,187 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.894e+02 3.789e+02 4.825e+02 7.842e+02, threshold=7.578e+02, percent-clipped=4.0 2022-12-07 13:29:52,492 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0322, 1.8268, 1.5337, 2.0844, 1.7448, 1.9808, 1.7164, 1.8140], device='cuda:1'), covar=tensor([0.0457, 0.1078, 0.2807, 0.0346, 0.0901, 0.0515, 0.1036, 0.0701], device='cuda:1'), in_proj_covar=tensor([0.0228, 0.0303, 0.0301, 0.0190, 0.0254, 0.0259, 0.0259, 0.0294], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:30:02,669 INFO [train.py:873] (1/4) Epoch 5, batch 6500, loss[loss=0.1816, simple_loss=0.1874, pruned_loss=0.08792, over 5982.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.1899, pruned_loss=0.08112, over 1969983.17 frames. ], batch size: 100, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:30:13,301 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:30:36,073 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5626, 2.4992, 2.4445, 2.2850, 1.9190, 2.6992, 2.4534, 0.9098], device='cuda:1'), covar=tensor([0.3467, 0.0978, 0.1303, 0.1678, 0.1311, 0.0641, 0.1592, 0.4669], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0061, 0.0053, 0.0054, 0.0074, 0.0057, 0.0083, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:30:51,583 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:31:05,624 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8871, 1.5386, 2.0295, 1.7547, 2.0110, 1.4880, 1.7371, 1.7606], device='cuda:1'), covar=tensor([0.0782, 0.1410, 0.0123, 0.1065, 0.0538, 0.0815, 0.0592, 0.0366], device='cuda:1'), in_proj_covar=tensor([0.0222, 0.0252, 0.0166, 0.0346, 0.0177, 0.0265, 0.0241, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:31:07,674 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.577e+02 3.368e+02 4.224e+02 7.324e+02, threshold=6.736e+02, percent-clipped=0.0 2022-12-07 13:31:18,652 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:31:28,830 INFO [train.py:873] (1/4) Epoch 5, batch 6600, loss[loss=0.1787, simple_loss=0.1839, pruned_loss=0.0867, over 5985.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.1902, pruned_loss=0.08115, over 2017307.62 frames. ], batch size: 100, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:31:44,143 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:32:00,561 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:32:23,274 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:32:34,637 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.557e+01 2.430e+02 3.121e+02 4.365e+02 9.299e+02, threshold=6.241e+02, percent-clipped=3.0 2022-12-07 13:32:35,615 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1077, 2.9374, 2.5213, 2.6656, 3.0089, 2.9156, 3.0838, 3.0343], device='cuda:1'), covar=tensor([0.1002, 0.0700, 0.2210, 0.3459, 0.0869, 0.1071, 0.1237, 0.1103], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0221, 0.0351, 0.0441, 0.0260, 0.0314, 0.0328, 0.0273], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:32:55,727 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.79 vs. limit=5.0 2022-12-07 13:32:56,984 INFO [train.py:873] (1/4) Epoch 5, batch 6700, loss[loss=0.1603, simple_loss=0.1897, pruned_loss=0.06544, over 14323.00 frames. ], tot_loss[loss=0.176, simple_loss=0.1898, pruned_loss=0.08112, over 1942984.33 frames. ], batch size: 31, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:32:58,055 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0774, 1.5172, 2.3735, 1.9326, 2.3330, 1.5743, 1.8906, 1.9969], device='cuda:1'), covar=tensor([0.1228, 0.3055, 0.0249, 0.2724, 0.0553, 0.2503, 0.1045, 0.0495], device='cuda:1'), in_proj_covar=tensor([0.0224, 0.0253, 0.0167, 0.0348, 0.0176, 0.0267, 0.0243, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:33:00,465 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2310, 1.9654, 2.2204, 1.3647, 1.7921, 2.0668, 2.2634, 1.9366], device='cuda:1'), covar=tensor([0.0575, 0.1066, 0.0916, 0.2101, 0.0882, 0.0853, 0.0405, 0.1556], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0203, 0.0117, 0.0126, 0.0102, 0.0110, 0.0088, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:33:04,727 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:07,489 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:10,002 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:13,446 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:16,861 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:20,293 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-07 13:33:30,698 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8998, 1.5330, 3.7103, 3.5246, 3.7304, 3.8054, 3.1397, 3.8721], device='cuda:1'), covar=tensor([0.1099, 0.1236, 0.0108, 0.0155, 0.0128, 0.0099, 0.0207, 0.0104], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0150, 0.0100, 0.0141, 0.0113, 0.0118, 0.0091, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 13:33:45,500 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:59,005 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:34:03,151 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.484e+02 3.120e+02 4.102e+02 1.074e+03, threshold=6.239e+02, percent-clipped=7.0 2022-12-07 13:34:24,643 INFO [train.py:873] (1/4) Epoch 5, batch 6800, loss[loss=0.1884, simple_loss=0.1874, pruned_loss=0.09475, over 4942.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.1889, pruned_loss=0.08005, over 1951778.86 frames. ], batch size: 100, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:34:31,284 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:34:38,930 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:35:07,257 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 13:35:30,147 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.403e+02 2.984e+02 3.959e+02 1.091e+03, threshold=5.968e+02, percent-clipped=8.0 2022-12-07 13:35:32,042 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0484, 1.4985, 3.2448, 1.6872, 3.0474, 3.1606, 2.2277, 3.3094], device='cuda:1'), covar=tensor([0.0202, 0.2414, 0.0237, 0.1811, 0.0648, 0.0300, 0.0707, 0.0174], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0161, 0.0138, 0.0168, 0.0156, 0.0150, 0.0122, 0.0122], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:35:52,586 INFO [train.py:873] (1/4) Epoch 5, batch 6900, loss[loss=0.1956, simple_loss=0.213, pruned_loss=0.08908, over 14466.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.1886, pruned_loss=0.07992, over 1997585.89 frames. ], batch size: 51, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:36:02,873 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:36:26,025 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4051, 3.2196, 2.9965, 3.0862, 3.3623, 3.2963, 3.3839, 3.3663], device='cuda:1'), covar=tensor([0.0900, 0.0663, 0.1785, 0.2788, 0.0756, 0.0781, 0.1122, 0.0973], device='cuda:1'), in_proj_covar=tensor([0.0304, 0.0233, 0.0371, 0.0469, 0.0275, 0.0325, 0.0343, 0.0284], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:36:58,808 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.434e+02 3.159e+02 4.022e+02 7.807e+02, threshold=6.317e+02, percent-clipped=6.0 2022-12-07 13:37:20,128 INFO [train.py:873] (1/4) Epoch 5, batch 7000, loss[loss=0.1992, simple_loss=0.1841, pruned_loss=0.1072, over 3876.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.1891, pruned_loss=0.08088, over 1917189.90 frames. ], batch size: 100, lr: 1.48e-02, grad_scale: 4.0 2022-12-07 13:37:31,299 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:33,780 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:36,214 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:37,158 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:37,194 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:37,506 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.44 vs. limit=5.0 2022-12-07 13:38:12,649 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:14,991 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:16,766 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:18,490 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:26,221 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.611e+02 3.349e+02 4.594e+02 9.259e+02, threshold=6.698e+02, percent-clipped=6.0 2022-12-07 13:38:26,453 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3882, 3.2207, 3.0173, 2.0089, 2.7053, 3.2112, 3.2788, 2.4504], device='cuda:1'), covar=tensor([0.0792, 0.2300, 0.1367, 0.2812, 0.1092, 0.0478, 0.1008, 0.2252], device='cuda:1'), in_proj_covar=tensor([0.0112, 0.0205, 0.0120, 0.0128, 0.0106, 0.0113, 0.0091, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 13:38:29,707 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:41,171 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9259, 1.6686, 2.0111, 1.8057, 2.1064, 1.8632, 1.6799, 2.0168], device='cuda:1'), covar=tensor([0.0190, 0.0826, 0.0100, 0.0217, 0.0105, 0.0184, 0.0089, 0.0152], device='cuda:1'), in_proj_covar=tensor([0.0273, 0.0334, 0.0364, 0.0312, 0.0342, 0.0282, 0.0330, 0.0350], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:38:46,859 INFO [train.py:873] (1/4) Epoch 5, batch 7100, loss[loss=0.1673, simple_loss=0.1897, pruned_loss=0.07248, over 14080.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.1896, pruned_loss=0.08085, over 1968101.50 frames. ], batch size: 26, lr: 1.48e-02, grad_scale: 4.0 2022-12-07 13:38:52,916 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:38:56,266 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:59,716 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:39:00,856 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9963, 3.9768, 4.2281, 3.5247, 4.0009, 4.0833, 1.5247, 3.8604], device='cuda:1'), covar=tensor([0.0177, 0.0265, 0.0310, 0.0448, 0.0255, 0.0227, 0.2823, 0.0242], device='cuda:1'), in_proj_covar=tensor([0.0129, 0.0135, 0.0123, 0.0114, 0.0169, 0.0116, 0.0149, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:39:12,509 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8369, 0.9961, 1.2357, 1.1526, 1.0112, 0.5248, 1.2289, 1.0310], device='cuda:1'), covar=tensor([0.0763, 0.0680, 0.0365, 0.1563, 0.1114, 0.0451, 0.0898, 0.0912], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0018, 0.0019, 0.0020, 0.0025, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([7.5546e-05, 8.0029e-05, 7.2509e-05, 7.7246e-05, 8.2974e-05, 9.5666e-05, 8.4872e-05, 7.8801e-05], device='cuda:1') 2022-12-07 13:39:24,420 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7043, 1.9300, 2.6850, 2.7418, 2.6997, 2.0242, 2.8487, 2.0605], device='cuda:1'), covar=tensor([0.0133, 0.0354, 0.0243, 0.0166, 0.0153, 0.0530, 0.0098, 0.0396], device='cuda:1'), in_proj_covar=tensor([0.0193, 0.0195, 0.0287, 0.0231, 0.0185, 0.0241, 0.0181, 0.0233], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 13:39:34,774 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:39:37,570 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2097, 4.8873, 4.5378, 4.7328, 4.6738, 5.0731, 5.1018, 5.1627], device='cuda:1'), covar=tensor([0.0654, 0.0419, 0.2256, 0.2568, 0.0849, 0.0596, 0.0998, 0.0663], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0228, 0.0372, 0.0464, 0.0275, 0.0331, 0.0342, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:39:53,448 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.303e+02 3.351e+02 4.171e+02 6.616e+02, threshold=6.701e+02, percent-clipped=0.0 2022-12-07 13:39:53,654 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:39:58,504 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 13:40:14,167 INFO [train.py:873] (1/4) Epoch 5, batch 7200, loss[loss=0.2135, simple_loss=0.21, pruned_loss=0.1085, over 10352.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.1907, pruned_loss=0.08271, over 1920167.70 frames. ], batch size: 100, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:40:25,505 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:40:46,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9425, 0.9473, 0.9482, 1.0475, 1.1289, 0.5145, 0.9884, 0.9311], device='cuda:1'), covar=tensor([0.0740, 0.0842, 0.1132, 0.0819, 0.0813, 0.0504, 0.0704, 0.0574], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0018, 0.0019, 0.0021, 0.0025, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([7.5856e-05, 8.0381e-05, 7.2735e-05, 7.7398e-05, 8.3531e-05, 9.6780e-05, 8.6047e-05, 7.9098e-05], device='cuda:1') 2022-12-07 13:40:51,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9350, 2.7870, 2.0560, 3.0179, 2.6780, 2.8146, 2.5452, 2.3190], device='cuda:1'), covar=tensor([0.0391, 0.1003, 0.3491, 0.0264, 0.0744, 0.0370, 0.1108, 0.2584], device='cuda:1'), in_proj_covar=tensor([0.0224, 0.0303, 0.0301, 0.0189, 0.0253, 0.0253, 0.0252, 0.0289], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:41:04,694 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4186, 4.1209, 3.8955, 3.9799, 4.1674, 4.2833, 4.4240, 4.3582], device='cuda:1'), covar=tensor([0.0965, 0.0622, 0.1924, 0.2814, 0.0792, 0.0766, 0.0941, 0.0883], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0231, 0.0372, 0.0462, 0.0277, 0.0331, 0.0343, 0.0283], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:41:08,039 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:41:21,487 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.612e+02 3.336e+02 4.094e+02 6.871e+02, threshold=6.672e+02, percent-clipped=1.0 2022-12-07 13:41:26,994 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2022-12-07 13:41:42,813 INFO [train.py:873] (1/4) Epoch 5, batch 7300, loss[loss=0.1782, simple_loss=0.1684, pruned_loss=0.09403, over 2658.00 frames. ], tot_loss[loss=0.177, simple_loss=0.1897, pruned_loss=0.08213, over 1890022.95 frames. ], batch size: 100, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:41:51,060 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2022-12-07 13:41:58,612 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:19,108 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7383, 3.0103, 4.1306, 2.5874, 2.4858, 2.9505, 1.5641, 3.1530], device='cuda:1'), covar=tensor([0.1425, 0.0619, 0.0338, 0.1923, 0.1710, 0.1125, 0.5030, 0.0900], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0080, 0.0078, 0.0086, 0.0109, 0.0071, 0.0141, 0.0071], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:42:27,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8806, 2.0040, 2.2331, 1.4128, 1.5722, 2.0077, 1.0691, 1.8970], device='cuda:1'), covar=tensor([0.1057, 0.1256, 0.0590, 0.2382, 0.2810, 0.0668, 0.5013, 0.0793], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0080, 0.0079, 0.0087, 0.0109, 0.0071, 0.0142, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:42:41,418 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:42,191 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:50,490 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:51,224 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.435e+02 3.251e+02 4.059e+02 7.905e+02, threshold=6.501e+02, percent-clipped=2.0 2022-12-07 13:43:12,151 INFO [train.py:873] (1/4) Epoch 5, batch 7400, loss[loss=0.1849, simple_loss=0.1978, pruned_loss=0.086, over 14209.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.1894, pruned_loss=0.0812, over 1957358.54 frames. ], batch size: 84, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:43:22,044 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:43:24,009 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:43:36,059 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2022-12-07 13:43:57,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8836, 2.8097, 4.8478, 3.3640, 4.6658, 2.1909, 3.7392, 4.6211], device='cuda:1'), covar=tensor([0.0304, 0.4085, 0.0240, 0.8126, 0.0275, 0.4340, 0.0904, 0.0228], device='cuda:1'), in_proj_covar=tensor([0.0226, 0.0255, 0.0169, 0.0352, 0.0178, 0.0272, 0.0243, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:43:59,418 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=12.31 vs. limit=5.0 2022-12-07 13:44:05,021 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:44:15,013 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:44:19,314 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.420e+02 3.401e+02 4.272e+02 7.052e+02, threshold=6.801e+02, percent-clipped=3.0 2022-12-07 13:44:41,184 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 13:44:41,303 INFO [train.py:873] (1/4) Epoch 5, batch 7500, loss[loss=0.181, simple_loss=0.1608, pruned_loss=0.1006, over 1217.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.1887, pruned_loss=0.08099, over 1901835.79 frames. ], batch size: 100, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:44:46,775 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8888, 1.2984, 1.4534, 1.3889, 1.1168, 0.5631, 1.4914, 1.4771], device='cuda:1'), covar=tensor([0.2346, 0.0787, 0.0420, 0.0678, 0.1925, 0.0467, 0.1087, 0.0935], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0017, 0.0018, 0.0020, 0.0025, 0.0020, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.5865e-05, 7.7856e-05, 7.0062e-05, 7.6076e-05, 8.1308e-05, 9.6031e-05, 8.5061e-05, 7.5130e-05], device='cuda:1') 2022-12-07 13:46:12,627 INFO [train.py:873] (1/4) Epoch 6, batch 0, loss[loss=0.2224, simple_loss=0.2257, pruned_loss=0.1096, over 14207.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.2257, pruned_loss=0.1096, over 14207.00 frames. ], batch size: 89, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:46:12,628 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 13:46:18,276 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9208, 1.7369, 2.9896, 1.8318, 3.0123, 2.9228, 2.1867, 3.0725], device='cuda:1'), covar=tensor([0.0189, 0.1148, 0.0221, 0.0848, 0.0371, 0.0300, 0.0807, 0.0156], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0154, 0.0132, 0.0164, 0.0151, 0.0146, 0.0119, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:46:20,010 INFO [train.py:905] (1/4) Epoch 6, validation: loss=0.1313, simple_loss=0.1749, pruned_loss=0.04388, over 857387.00 frames. 2022-12-07 13:46:20,010 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 13:46:32,651 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.323e+01 1.772e+02 2.799e+02 3.695e+02 8.641e+02, threshold=5.598e+02, percent-clipped=1.0 2022-12-07 13:46:43,453 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 13:47:34,033 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8416, 2.2053, 2.0884, 2.0729, 1.7377, 2.1913, 1.8466, 1.0792], device='cuda:1'), covar=tensor([0.2014, 0.0626, 0.0893, 0.0571, 0.0959, 0.0454, 0.1382, 0.3080], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0059, 0.0051, 0.0053, 0.0073, 0.0056, 0.0080, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:47:36,095 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.55 vs. limit=5.0 2022-12-07 13:47:50,544 INFO [train.py:873] (1/4) Epoch 6, batch 100, loss[loss=0.1729, simple_loss=0.1912, pruned_loss=0.07728, over 14430.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.1917, pruned_loss=0.08152, over 852859.84 frames. ], batch size: 53, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:48:01,934 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:48:02,650 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.495e+02 3.181e+02 4.062e+02 9.836e+02, threshold=6.363e+02, percent-clipped=3.0 2022-12-07 13:48:42,073 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4609, 3.0937, 2.7545, 2.0624, 2.8146, 2.9967, 3.1683, 2.5794], device='cuda:1'), covar=tensor([0.0545, 0.3145, 0.1253, 0.2672, 0.0964, 0.0669, 0.1290, 0.1704], device='cuda:1'), in_proj_covar=tensor([0.0107, 0.0199, 0.0118, 0.0124, 0.0104, 0.0110, 0.0088, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 13:48:44,519 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:48:50,758 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:49:16,101 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7212, 1.3149, 1.3569, 1.3151, 1.0610, 1.3534, 1.0004, 0.8668], device='cuda:1'), covar=tensor([0.2601, 0.0454, 0.0289, 0.0374, 0.0941, 0.0302, 0.1041, 0.1059], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0058, 0.0051, 0.0052, 0.0073, 0.0056, 0.0079, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:49:19,531 INFO [train.py:873] (1/4) Epoch 6, batch 200, loss[loss=0.2127, simple_loss=0.2102, pruned_loss=0.1076, over 8626.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1916, pruned_loss=0.08302, over 1256766.70 frames. ], batch size: 100, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:49:27,114 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:49:31,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.443e+02 3.000e+02 3.883e+02 6.442e+02, threshold=6.000e+02, percent-clipped=2.0 2022-12-07 13:49:43,940 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8317, 2.6577, 4.8657, 3.0882, 4.6495, 2.1510, 3.5346, 4.4905], device='cuda:1'), covar=tensor([0.0251, 0.4427, 0.0195, 0.8837, 0.0270, 0.3987, 0.1093, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0254, 0.0165, 0.0346, 0.0177, 0.0266, 0.0242, 0.0175], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:49:45,668 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:50:09,553 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:50:23,242 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1972, 2.2001, 2.8280, 1.9346, 1.9847, 2.5661, 1.2026, 2.2797], device='cuda:1'), covar=tensor([0.1412, 0.1842, 0.1088, 0.2094, 0.2761, 0.1085, 0.7457, 0.1578], device='cuda:1'), in_proj_covar=tensor([0.0070, 0.0080, 0.0076, 0.0086, 0.0109, 0.0069, 0.0136, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:50:47,795 INFO [train.py:873] (1/4) Epoch 6, batch 300, loss[loss=0.1633, simple_loss=0.1857, pruned_loss=0.07046, over 14227.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.1903, pruned_loss=0.08266, over 1490146.86 frames. ], batch size: 60, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:50:59,800 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.428e+02 3.008e+02 3.765e+02 6.349e+02, threshold=6.017e+02, percent-clipped=1.0 2022-12-07 13:52:16,131 INFO [train.py:873] (1/4) Epoch 6, batch 400, loss[loss=0.172, simple_loss=0.1871, pruned_loss=0.07847, over 14230.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.1888, pruned_loss=0.0814, over 1640520.75 frames. ], batch size: 69, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:52:18,536 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 13:52:25,576 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 13:52:28,766 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.719e+02 3.194e+02 3.868e+02 8.474e+02, threshold=6.387e+02, percent-clipped=5.0 2022-12-07 13:52:42,715 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.34 vs. limit=5.0 2022-12-07 13:53:00,585 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5276, 3.3630, 2.6982, 1.9130, 3.1202, 3.5306, 3.3211, 2.5744], device='cuda:1'), covar=tensor([0.0705, 0.3503, 0.1443, 0.3206, 0.0895, 0.0516, 0.1459, 0.1964], device='cuda:1'), in_proj_covar=tensor([0.0109, 0.0200, 0.0116, 0.0126, 0.0104, 0.0109, 0.0088, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 13:53:04,519 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.19 vs. limit=2.0 2022-12-07 13:53:25,842 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0665, 2.0621, 3.1644, 3.1795, 3.0270, 2.3255, 3.1344, 2.5349], device='cuda:1'), covar=tensor([0.0120, 0.0308, 0.0235, 0.0170, 0.0118, 0.0419, 0.0086, 0.0352], device='cuda:1'), in_proj_covar=tensor([0.0200, 0.0200, 0.0297, 0.0238, 0.0190, 0.0246, 0.0181, 0.0238], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 13:53:29,933 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0878, 2.2669, 5.0538, 4.6866, 4.3664, 5.0818, 4.9074, 5.0550], device='cuda:1'), covar=tensor([0.1318, 0.1242, 0.0074, 0.0087, 0.0137, 0.0081, 0.0055, 0.0095], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0155, 0.0104, 0.0142, 0.0118, 0.0122, 0.0094, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 13:53:45,490 INFO [train.py:873] (1/4) Epoch 6, batch 500, loss[loss=0.1985, simple_loss=0.1814, pruned_loss=0.1078, over 3895.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.1879, pruned_loss=0.08022, over 1704817.53 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:53:57,615 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.649e+02 3.367e+02 4.608e+02 8.885e+02, threshold=6.735e+02, percent-clipped=8.0 2022-12-07 13:54:06,530 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:54:18,102 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8293, 2.0063, 3.8078, 2.7065, 3.8333, 1.9325, 2.8808, 3.7229], device='cuda:1'), covar=tensor([0.0448, 0.5203, 0.0359, 0.8866, 0.0389, 0.4251, 0.1447, 0.0326], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0253, 0.0168, 0.0347, 0.0178, 0.0262, 0.0238, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 13:54:27,919 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8699, 1.4329, 1.7863, 1.7682, 2.0354, 1.7507, 1.5981, 1.8611], device='cuda:1'), covar=tensor([0.0272, 0.1022, 0.0178, 0.0328, 0.0169, 0.0478, 0.0209, 0.0331], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0337, 0.0367, 0.0317, 0.0350, 0.0290, 0.0337, 0.0349], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:54:28,802 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:55:12,571 INFO [train.py:873] (1/4) Epoch 6, batch 600, loss[loss=0.2008, simple_loss=0.2057, pruned_loss=0.09793, over 7815.00 frames. ], tot_loss[loss=0.174, simple_loss=0.1877, pruned_loss=0.08011, over 1825821.62 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:55:21,335 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:55:24,768 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.417e+02 3.018e+02 3.905e+02 9.436e+02, threshold=6.036e+02, percent-clipped=5.0 2022-12-07 13:55:24,931 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4694, 0.9375, 1.3964, 0.8962, 1.2081, 1.3294, 1.1471, 1.1333], device='cuda:1'), covar=tensor([0.0374, 0.0847, 0.0448, 0.0453, 0.0749, 0.0561, 0.0296, 0.1068], device='cuda:1'), in_proj_covar=tensor([0.0106, 0.0197, 0.0114, 0.0122, 0.0102, 0.0109, 0.0087, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 13:55:45,812 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6922, 0.6542, 0.5513, 0.6921, 0.7523, 0.2004, 0.4863, 0.6194], device='cuda:1'), covar=tensor([0.0211, 0.0278, 0.0122, 0.0209, 0.0175, 0.0129, 0.0308, 0.0403], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0020, 0.0026, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([7.8789e-05, 8.2121e-05, 7.2772e-05, 7.9142e-05, 8.4275e-05, 1.0033e-04, 8.6995e-05, 7.8960e-05], device='cuda:1') 2022-12-07 13:56:30,493 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2022-12-07 13:56:41,309 INFO [train.py:873] (1/4) Epoch 6, batch 700, loss[loss=0.2443, simple_loss=0.1985, pruned_loss=0.1451, over 1224.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.1871, pruned_loss=0.08015, over 1766266.56 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:56:53,909 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.144e+02 2.377e+02 3.088e+02 3.950e+02 6.999e+02, threshold=6.177e+02, percent-clipped=2.0 2022-12-07 13:58:10,603 INFO [train.py:873] (1/4) Epoch 6, batch 800, loss[loss=0.1635, simple_loss=0.1506, pruned_loss=0.08821, over 1290.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.1869, pruned_loss=0.0797, over 1861485.87 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:58:13,810 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 13:58:19,522 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6690, 2.2851, 3.8023, 3.8867, 3.7991, 2.3405, 3.8332, 2.9959], device='cuda:1'), covar=tensor([0.0140, 0.0400, 0.0287, 0.0175, 0.0112, 0.0567, 0.0104, 0.0401], device='cuda:1'), in_proj_covar=tensor([0.0195, 0.0194, 0.0289, 0.0233, 0.0184, 0.0240, 0.0177, 0.0232], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 13:58:23,024 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.424e+02 2.905e+02 3.924e+02 6.932e+02, threshold=5.809e+02, percent-clipped=2.0 2022-12-07 13:58:32,272 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:58:33,582 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 13:59:15,085 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:59:37,890 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4161, 2.3471, 2.4190, 2.4409, 2.3711, 2.1629, 1.3386, 2.1645], device='cuda:1'), covar=tensor([0.0395, 0.0458, 0.0499, 0.0345, 0.0322, 0.0754, 0.2456, 0.0375], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0139, 0.0127, 0.0119, 0.0170, 0.0118, 0.0155, 0.0159], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 13:59:39,509 INFO [train.py:873] (1/4) Epoch 6, batch 900, loss[loss=0.1719, simple_loss=0.1865, pruned_loss=0.0786, over 12738.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.1871, pruned_loss=0.07925, over 1912347.70 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:59:44,623 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:59:52,235 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.457e+02 3.272e+02 3.954e+02 1.451e+03, threshold=6.544e+02, percent-clipped=7.0 2022-12-07 14:00:05,625 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2240, 2.1185, 3.2137, 3.3293, 3.3538, 2.2335, 3.2544, 2.5371], device='cuda:1'), covar=tensor([0.0152, 0.0340, 0.0337, 0.0169, 0.0131, 0.0547, 0.0123, 0.0440], device='cuda:1'), in_proj_covar=tensor([0.0197, 0.0197, 0.0295, 0.0235, 0.0186, 0.0241, 0.0181, 0.0235], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:00:17,155 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5215, 3.2624, 2.8627, 1.8353, 2.9198, 3.1113, 3.5872, 2.6211], device='cuda:1'), covar=tensor([0.0607, 0.2324, 0.1246, 0.2715, 0.0932, 0.0701, 0.0801, 0.1681], device='cuda:1'), in_proj_covar=tensor([0.0111, 0.0206, 0.0120, 0.0128, 0.0109, 0.0115, 0.0089, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:00:44,547 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1094, 3.1158, 3.0657, 2.9887, 2.9934, 2.4461, 1.2409, 2.8634], device='cuda:1'), covar=tensor([0.0341, 0.0442, 0.0691, 0.0592, 0.0512, 0.1218, 0.4078, 0.0393], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0135, 0.0124, 0.0116, 0.0166, 0.0118, 0.0152, 0.0156], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:01:08,901 INFO [train.py:873] (1/4) Epoch 6, batch 1000, loss[loss=0.1613, simple_loss=0.1832, pruned_loss=0.0697, over 14191.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.1875, pruned_loss=0.07931, over 1888995.83 frames. ], batch size: 94, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:01:19,464 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:01:21,141 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 2.592e+02 3.216e+02 4.248e+02 8.098e+02, threshold=6.432e+02, percent-clipped=2.0 2022-12-07 14:02:06,279 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5120, 3.2186, 2.7636, 1.6617, 2.8331, 3.2394, 3.5337, 2.5402], device='cuda:1'), covar=tensor([0.0600, 0.2196, 0.1197, 0.2773, 0.1070, 0.0547, 0.0649, 0.1632], device='cuda:1'), in_proj_covar=tensor([0.0110, 0.0204, 0.0118, 0.0126, 0.0108, 0.0113, 0.0087, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:02:13,797 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:02:14,236 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 14:02:32,899 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0800, 2.9174, 4.0047, 2.5994, 2.4562, 3.0859, 1.8199, 3.2664], device='cuda:1'), covar=tensor([0.1294, 0.0738, 0.0376, 0.2041, 0.2166, 0.1020, 0.4520, 0.0876], device='cuda:1'), in_proj_covar=tensor([0.0067, 0.0073, 0.0073, 0.0082, 0.0102, 0.0066, 0.0127, 0.0070], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 14:02:37,251 INFO [train.py:873] (1/4) Epoch 6, batch 1100, loss[loss=0.1623, simple_loss=0.1816, pruned_loss=0.07149, over 14171.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.1874, pruned_loss=0.07906, over 1927022.69 frames. ], batch size: 99, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:02:46,901 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9123, 1.4082, 3.1649, 1.4869, 2.9500, 3.0597, 2.2881, 3.2587], device='cuda:1'), covar=tensor([0.0269, 0.2532, 0.0286, 0.2177, 0.0770, 0.0373, 0.0755, 0.0190], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0160, 0.0141, 0.0173, 0.0157, 0.0153, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:02:50,186 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.088e+02 2.398e+02 3.098e+02 3.888e+02 7.053e+02, threshold=6.197e+02, percent-clipped=1.0 2022-12-07 14:03:14,157 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.77 vs. limit=5.0 2022-12-07 14:03:58,432 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8796, 4.5769, 4.2917, 4.3283, 4.4829, 4.6425, 4.9142, 4.8251], device='cuda:1'), covar=tensor([0.0930, 0.0543, 0.1977, 0.3372, 0.0649, 0.0699, 0.0810, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0303, 0.0232, 0.0371, 0.0464, 0.0272, 0.0337, 0.0342, 0.0283], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:04:07,243 INFO [train.py:873] (1/4) Epoch 6, batch 1200, loss[loss=0.2551, simple_loss=0.2291, pruned_loss=0.1405, over 7794.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.1869, pruned_loss=0.07817, over 1963128.68 frames. ], batch size: 100, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:04:11,619 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:04:19,444 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.136e+01 2.396e+02 3.010e+02 3.930e+02 7.519e+02, threshold=6.020e+02, percent-clipped=3.0 2022-12-07 14:04:54,205 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:04:59,333 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 14:05:13,388 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1896, 1.8150, 2.2140, 1.3729, 1.7635, 2.0946, 2.1767, 1.7736], device='cuda:1'), covar=tensor([0.0618, 0.1040, 0.0976, 0.2200, 0.1082, 0.0477, 0.0416, 0.1767], device='cuda:1'), in_proj_covar=tensor([0.0108, 0.0201, 0.0117, 0.0125, 0.0106, 0.0111, 0.0088, 0.0126], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:05:35,513 INFO [train.py:873] (1/4) Epoch 6, batch 1300, loss[loss=0.1833, simple_loss=0.1948, pruned_loss=0.08584, over 12006.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.1865, pruned_loss=0.07781, over 1978308.14 frames. ], batch size: 100, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:05:48,566 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.179e+02 2.329e+02 2.773e+02 3.737e+02 7.179e+02, threshold=5.547e+02, percent-clipped=2.0 2022-12-07 14:05:57,257 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0215, 1.4822, 3.9454, 1.7172, 3.8708, 4.0531, 3.1797, 4.3338], device='cuda:1'), covar=tensor([0.0244, 0.3372, 0.0426, 0.2808, 0.0383, 0.0384, 0.0647, 0.0214], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0159, 0.0139, 0.0172, 0.0155, 0.0152, 0.0124, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:06:36,465 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:06:41,863 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:06:53,007 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-07 14:07:05,238 INFO [train.py:873] (1/4) Epoch 6, batch 1400, loss[loss=0.1617, simple_loss=0.1873, pruned_loss=0.0681, over 14270.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.1867, pruned_loss=0.07817, over 1939753.02 frames. ], batch size: 76, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:07:17,497 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.370e+02 2.964e+02 3.794e+02 7.798e+02, threshold=5.929e+02, percent-clipped=9.0 2022-12-07 14:07:31,918 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-07 14:07:36,779 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:07:58,729 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.87 vs. limit=5.0 2022-12-07 14:08:10,687 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:08:15,483 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.63 vs. limit=5.0 2022-12-07 14:08:26,465 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9579, 1.9657, 5.0355, 4.5995, 4.5311, 5.0910, 4.9402, 5.0903], device='cuda:1'), covar=tensor([0.1224, 0.1146, 0.0054, 0.0111, 0.0105, 0.0065, 0.0058, 0.0084], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0153, 0.0105, 0.0144, 0.0119, 0.0124, 0.0095, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 14:08:33,849 INFO [train.py:873] (1/4) Epoch 6, batch 1500, loss[loss=0.2023, simple_loss=0.1698, pruned_loss=0.1174, over 1223.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.1859, pruned_loss=0.07759, over 1962420.13 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 16.0 2022-12-07 14:08:34,419 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 14:08:38,161 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5211, 2.3892, 1.8928, 2.5466, 2.2609, 2.4078, 2.1910, 2.0515], device='cuda:1'), covar=tensor([0.0375, 0.0880, 0.2045, 0.0276, 0.1021, 0.0430, 0.1012, 0.1281], device='cuda:1'), in_proj_covar=tensor([0.0226, 0.0301, 0.0296, 0.0189, 0.0256, 0.0252, 0.0257, 0.0288], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:08:38,981 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1929, 1.3259, 3.3708, 1.4383, 3.0561, 3.2275, 2.4865, 3.5091], device='cuda:1'), covar=tensor([0.0176, 0.2564, 0.0252, 0.1987, 0.0805, 0.0385, 0.0708, 0.0148], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0159, 0.0137, 0.0171, 0.0155, 0.0151, 0.0123, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:08:46,905 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.464e+02 2.970e+02 3.813e+02 8.133e+02, threshold=5.940e+02, percent-clipped=4.0 2022-12-07 14:09:04,277 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:10:03,120 INFO [train.py:873] (1/4) Epoch 6, batch 1600, loss[loss=0.1445, simple_loss=0.1451, pruned_loss=0.07191, over 3865.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.1856, pruned_loss=0.07726, over 1970794.59 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:10:16,105 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.549e+01 2.367e+02 3.070e+02 3.991e+02 2.269e+03, threshold=6.141e+02, percent-clipped=9.0 2022-12-07 14:10:49,272 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:03,599 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:13,637 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6908, 4.3987, 4.2980, 4.6936, 4.2020, 3.9301, 4.7216, 4.6850], device='cuda:1'), covar=tensor([0.0545, 0.0541, 0.0502, 0.0485, 0.0598, 0.0473, 0.0493, 0.0516], device='cuda:1'), in_proj_covar=tensor([0.0112, 0.0098, 0.0113, 0.0115, 0.0116, 0.0088, 0.0128, 0.0109], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 14:11:31,582 INFO [train.py:873] (1/4) Epoch 6, batch 1700, loss[loss=0.1856, simple_loss=0.2013, pruned_loss=0.08495, over 14477.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.1868, pruned_loss=0.07829, over 1972695.45 frames. ], batch size: 24, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:11:33,922 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2022-12-07 14:11:44,039 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:45,579 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.465e+02 3.198e+02 4.096e+02 7.326e+02, threshold=6.396e+02, percent-clipped=2.0 2022-12-07 14:11:46,571 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:58,780 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:13:01,696 INFO [train.py:873] (1/4) Epoch 6, batch 1800, loss[loss=0.1799, simple_loss=0.1994, pruned_loss=0.0802, over 14267.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.1862, pruned_loss=0.07815, over 1898706.87 frames. ], batch size: 76, lr: 1.34e-02, grad_scale: 4.0 2022-12-07 14:13:15,825 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.436e+02 3.001e+02 3.832e+02 8.283e+02, threshold=6.003e+02, percent-clipped=3.0 2022-12-07 14:13:26,764 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:13:27,552 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:14:20,093 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:14:29,790 INFO [train.py:873] (1/4) Epoch 6, batch 1900, loss[loss=0.1535, simple_loss=0.1805, pruned_loss=0.06323, over 14036.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.1861, pruned_loss=0.07784, over 1979350.28 frames. ], batch size: 26, lr: 1.34e-02, grad_scale: 4.0 2022-12-07 14:14:44,080 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.151e+02 2.287e+02 3.025e+02 3.865e+02 4.213e+03, threshold=6.050e+02, percent-clipped=8.0 2022-12-07 14:15:15,863 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1316, 2.0633, 3.2985, 3.4260, 3.2175, 2.2052, 3.4845, 2.2813], device='cuda:1'), covar=tensor([0.0185, 0.0476, 0.0430, 0.0210, 0.0168, 0.0697, 0.0112, 0.0627], device='cuda:1'), in_proj_covar=tensor([0.0203, 0.0203, 0.0300, 0.0239, 0.0188, 0.0248, 0.0188, 0.0238], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:15:19,650 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:15:58,845 INFO [train.py:873] (1/4) Epoch 6, batch 2000, loss[loss=0.1597, simple_loss=0.1814, pruned_loss=0.06896, over 14359.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.1871, pruned_loss=0.0784, over 1969983.55 frames. ], batch size: 31, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:16:02,586 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:05,912 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:06,018 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9674, 1.5985, 1.8876, 1.7866, 2.1272, 1.8581, 1.6589, 1.8862], device='cuda:1'), covar=tensor([0.0189, 0.0957, 0.0164, 0.0277, 0.0144, 0.0287, 0.0148, 0.0237], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0337, 0.0375, 0.0311, 0.0355, 0.0290, 0.0342, 0.0352], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 14:16:12,945 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.241e+02 2.938e+02 3.748e+02 1.021e+03, threshold=5.875e+02, percent-clipped=4.0 2022-12-07 14:16:14,091 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:23,276 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.49 vs. limit=5.0 2022-12-07 14:16:25,432 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:47,995 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2022-12-07 14:16:56,529 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:17:02,305 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4548, 2.9401, 4.0659, 3.1521, 4.1007, 4.1337, 3.7972, 3.5381], device='cuda:1'), covar=tensor([0.0298, 0.2755, 0.0825, 0.1839, 0.0713, 0.0587, 0.1787, 0.1942], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0333, 0.0374, 0.0309, 0.0357, 0.0289, 0.0341, 0.0350], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 14:17:08,341 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:17:27,453 INFO [train.py:873] (1/4) Epoch 6, batch 2100, loss[loss=0.1834, simple_loss=0.1729, pruned_loss=0.09696, over 2688.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.1862, pruned_loss=0.07711, over 1999693.69 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:17:42,097 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.442e+01 2.310e+02 2.903e+02 3.569e+02 8.046e+02, threshold=5.805e+02, percent-clipped=4.0 2022-12-07 14:17:53,582 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:17:54,133 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 14:18:35,846 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:18:41,887 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 14:19:00,410 INFO [train.py:873] (1/4) Epoch 6, batch 2200, loss[loss=0.2059, simple_loss=0.1945, pruned_loss=0.1087, over 4963.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.1868, pruned_loss=0.0779, over 2021413.18 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:19:14,065 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.175e+01 2.567e+02 3.218e+02 4.341e+02 1.607e+03, threshold=6.436e+02, percent-clipped=13.0 2022-12-07 14:19:37,005 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 14:20:18,903 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0693, 1.2384, 1.3020, 0.7827, 0.8483, 0.9593, 0.9891, 0.6632], device='cuda:1'), covar=tensor([0.0340, 0.0382, 0.0463, 0.0327, 0.0656, 0.0406, 0.0288, 0.0774], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0017], device='cuda:1'), out_proj_covar=tensor([6.4423e-05, 6.6874e-05, 6.3704e-05, 6.2691e-05, 6.8888e-05, 9.1040e-05, 7.5962e-05, 8.9197e-05], device='cuda:1') 2022-12-07 14:20:27,390 INFO [train.py:873] (1/4) Epoch 6, batch 2300, loss[loss=0.156, simple_loss=0.178, pruned_loss=0.06702, over 14217.00 frames. ], tot_loss[loss=0.17, simple_loss=0.1857, pruned_loss=0.0771, over 1991140.91 frames. ], batch size: 35, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:20:27,970 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 14:20:31,325 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 14:20:35,062 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:20:38,462 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:20:41,820 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.659e+01 2.460e+02 3.103e+02 4.082e+02 7.462e+02, threshold=6.206e+02, percent-clipped=2.0 2022-12-07 14:21:18,097 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:18,452 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 14:21:20,036 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1468, 2.6806, 3.8122, 3.0599, 3.8172, 3.7721, 3.6930, 3.2655], device='cuda:1'), covar=tensor([0.0372, 0.2721, 0.0681, 0.1703, 0.0612, 0.0605, 0.1519, 0.1869], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0335, 0.0373, 0.0312, 0.0352, 0.0290, 0.0338, 0.0347], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 14:21:21,666 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:42,947 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:46,056 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 14:21:57,513 INFO [train.py:873] (1/4) Epoch 6, batch 2400, loss[loss=0.1836, simple_loss=0.1979, pruned_loss=0.08468, over 14257.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.1859, pruned_loss=0.07752, over 1976697.54 frames. ], batch size: 57, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:22:03,758 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:05,897 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 14:22:08,558 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-07 14:22:11,491 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.289e+02 3.018e+02 4.008e+02 1.305e+03, threshold=6.036e+02, percent-clipped=3.0 2022-12-07 14:22:37,337 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:51,150 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:57,613 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:04,725 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7308, 1.9912, 2.3284, 1.4067, 1.6435, 2.0605, 1.2488, 1.9738], device='cuda:1'), covar=tensor([0.1478, 0.1405, 0.0571, 0.2152, 0.2511, 0.0692, 0.4368, 0.0821], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0077, 0.0076, 0.0084, 0.0106, 0.0069, 0.0132, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:23:11,432 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:23:20,409 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0100, 5.0050, 5.3707, 4.5465, 5.0365, 5.5804, 2.0731, 4.8154], device='cuda:1'), covar=tensor([0.0148, 0.0217, 0.0263, 0.0319, 0.0210, 0.0085, 0.2689, 0.0184], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0136, 0.0122, 0.0116, 0.0169, 0.0117, 0.0150, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:23:25,880 INFO [train.py:873] (1/4) Epoch 6, batch 2500, loss[loss=0.1585, simple_loss=0.1771, pruned_loss=0.06991, over 14256.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.1863, pruned_loss=0.07797, over 1986660.88 frames. ], batch size: 80, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:23:39,898 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.352e+02 2.913e+02 3.857e+02 7.423e+02, threshold=5.826e+02, percent-clipped=3.0 2022-12-07 14:23:40,926 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-07 14:23:42,314 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7986, 2.3382, 3.8198, 3.9829, 3.9511, 2.5382, 4.0319, 3.2873], device='cuda:1'), covar=tensor([0.0164, 0.0409, 0.0535, 0.0196, 0.0121, 0.0593, 0.0102, 0.0393], device='cuda:1'), in_proj_covar=tensor([0.0203, 0.0206, 0.0301, 0.0241, 0.0190, 0.0252, 0.0190, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:23:43,145 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:45,720 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:49,608 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 14:23:53,917 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:24:36,857 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:24:54,952 INFO [train.py:873] (1/4) Epoch 6, batch 2600, loss[loss=0.1522, simple_loss=0.1769, pruned_loss=0.06381, over 14229.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.1849, pruned_loss=0.07641, over 2013949.59 frames. ], batch size: 57, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:25:05,698 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:25:08,968 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.477e+02 3.292e+02 4.100e+02 6.300e+02, threshold=6.583e+02, percent-clipped=3.0 2022-12-07 14:25:25,986 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8726, 1.9300, 2.2396, 1.2340, 1.5244, 1.9650, 1.1997, 1.8447], device='cuda:1'), covar=tensor([0.1137, 0.1497, 0.0513, 0.2601, 0.2758, 0.0840, 0.4122, 0.0928], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0078, 0.0076, 0.0085, 0.0105, 0.0069, 0.0132, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:25:36,328 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 14:25:47,485 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:25:47,565 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:06,956 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:23,202 INFO [train.py:873] (1/4) Epoch 6, batch 2700, loss[loss=0.1493, simple_loss=0.149, pruned_loss=0.07484, over 2620.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.1852, pruned_loss=0.07654, over 2006492.63 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:26:25,461 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 14:26:30,857 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:37,692 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.565e+02 3.016e+02 3.792e+02 1.283e+03, threshold=6.032e+02, percent-clipped=5.0 2022-12-07 14:26:58,592 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:00,641 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:19,400 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:51,008 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 14:27:52,085 INFO [train.py:873] (1/4) Epoch 6, batch 2800, loss[loss=0.1654, simple_loss=0.1697, pruned_loss=0.08055, over 4951.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.1853, pruned_loss=0.07584, over 2079600.88 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:28:05,952 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.405e+02 3.321e+02 4.166e+02 7.949e+02, threshold=6.642e+02, percent-clipped=7.0 2022-12-07 14:28:06,914 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:28:08,104 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-07 14:28:16,348 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-07 14:28:20,975 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1145, 2.8901, 2.1862, 3.0788, 2.9248, 2.9987, 2.6114, 2.3635], device='cuda:1'), covar=tensor([0.0508, 0.1176, 0.3369, 0.0297, 0.0677, 0.0627, 0.1307, 0.3141], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0294, 0.0289, 0.0187, 0.0255, 0.0254, 0.0257, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:28:24,912 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 14:28:27,751 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:28:29,797 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2022-12-07 14:28:58,729 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:29:20,280 INFO [train.py:873] (1/4) Epoch 6, batch 2900, loss[loss=0.1959, simple_loss=0.2008, pruned_loss=0.09547, over 12791.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.1851, pruned_loss=0.07588, over 2019794.70 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:29:21,320 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:29:23,388 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6811, 3.5630, 3.8933, 3.5284, 3.6123, 3.5711, 1.3811, 3.4886], device='cuda:1'), covar=tensor([0.0217, 0.0335, 0.0285, 0.0300, 0.0312, 0.0370, 0.3034, 0.0241], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0134, 0.0122, 0.0114, 0.0166, 0.0116, 0.0147, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:29:31,464 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1025, 2.4902, 3.9214, 4.1845, 4.1930, 2.6034, 4.3163, 3.3062], device='cuda:1'), covar=tensor([0.0132, 0.0460, 0.0368, 0.0177, 0.0145, 0.0637, 0.0088, 0.0471], device='cuda:1'), in_proj_covar=tensor([0.0205, 0.0205, 0.0308, 0.0243, 0.0194, 0.0253, 0.0195, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:29:34,595 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.325e+02 2.437e+02 2.821e+02 3.804e+02 6.190e+02, threshold=5.643e+02, percent-clipped=0.0 2022-12-07 14:29:42,334 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9042, 2.6222, 3.5096, 2.2328, 2.1345, 2.7965, 1.5941, 3.0352], device='cuda:1'), covar=tensor([0.1027, 0.0966, 0.0689, 0.3347, 0.2320, 0.0751, 0.4631, 0.0931], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0077, 0.0075, 0.0082, 0.0104, 0.0067, 0.0131, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:30:29,531 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 14:30:48,711 INFO [train.py:873] (1/4) Epoch 6, batch 3000, loss[loss=0.1901, simple_loss=0.1942, pruned_loss=0.09303, over 14542.00 frames. ], tot_loss[loss=0.168, simple_loss=0.1849, pruned_loss=0.07553, over 2058753.24 frames. ], batch size: 34, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:30:48,711 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 14:31:04,490 INFO [train.py:905] (1/4) Epoch 6, validation: loss=0.1224, simple_loss=0.1659, pruned_loss=0.03945, over 857387.00 frames. 2022-12-07 14:31:04,490 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 14:31:13,315 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 14:31:19,219 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.206e+02 2.431e+02 3.272e+02 4.116e+02 8.676e+02, threshold=6.543e+02, percent-clipped=8.0 2022-12-07 14:31:37,465 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:31:39,002 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 14:31:40,155 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:31:53,091 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:00,790 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:22,700 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:23,731 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:33,300 INFO [train.py:873] (1/4) Epoch 6, batch 3100, loss[loss=0.17, simple_loss=0.1809, pruned_loss=0.07952, over 14244.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.1856, pruned_loss=0.07626, over 2065034.94 frames. ], batch size: 69, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:32:43,004 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:46,558 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:47,206 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.510e+02 3.195e+02 3.907e+02 1.074e+03, threshold=6.390e+02, percent-clipped=2.0 2022-12-07 14:32:48,235 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:13,591 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 14:33:17,786 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:33:30,781 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:40,186 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:43,645 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2960, 2.7905, 2.7729, 1.9770, 2.6821, 3.1044, 3.0249, 2.5591], device='cuda:1'), covar=tensor([0.0506, 0.2307, 0.1147, 0.2166, 0.1000, 0.0615, 0.0863, 0.1453], device='cuda:1'), in_proj_covar=tensor([0.0111, 0.0199, 0.0121, 0.0126, 0.0106, 0.0111, 0.0088, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:33:49,420 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:58,623 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:01,827 INFO [train.py:873] (1/4) Epoch 6, batch 3200, loss[loss=0.1391, simple_loss=0.1769, pruned_loss=0.05066, over 14067.00 frames. ], tot_loss[loss=0.169, simple_loss=0.1854, pruned_loss=0.07635, over 2015828.55 frames. ], batch size: 22, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:34:16,163 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.950e+01 2.679e+02 3.334e+02 4.339e+02 1.581e+03, threshold=6.667e+02, percent-clipped=5.0 2022-12-07 14:34:22,471 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:32,028 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:39,364 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4488, 1.0061, 1.4024, 0.9876, 1.0804, 1.3498, 1.1227, 1.2164], device='cuda:1'), covar=tensor([0.0316, 0.0593, 0.0416, 0.0326, 0.0551, 0.0630, 0.0332, 0.0894], device='cuda:1'), in_proj_covar=tensor([0.0114, 0.0202, 0.0123, 0.0128, 0.0107, 0.0114, 0.0090, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:34:43,629 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:56,948 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1804, 2.8195, 2.2230, 3.2323, 2.9780, 3.0551, 2.5533, 2.3404], device='cuda:1'), covar=tensor([0.0477, 0.1399, 0.3772, 0.0407, 0.0828, 0.0935, 0.1531, 0.3876], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0306, 0.0297, 0.0195, 0.0262, 0.0263, 0.0263, 0.0287], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:35:22,834 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:35:26,834 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:35:30,960 INFO [train.py:873] (1/4) Epoch 6, batch 3300, loss[loss=0.144, simple_loss=0.1491, pruned_loss=0.06947, over 2578.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.1851, pruned_loss=0.07635, over 2011568.51 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:35:45,225 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.460e+02 3.115e+02 3.768e+02 7.967e+02, threshold=6.230e+02, percent-clipped=3.0 2022-12-07 14:36:04,013 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:17,081 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:46,073 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:59,351 INFO [train.py:873] (1/4) Epoch 6, batch 3400, loss[loss=0.161, simple_loss=0.1461, pruned_loss=0.08794, over 1244.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.1852, pruned_loss=0.07681, over 1946400.75 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:37:08,214 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:37:13,996 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.427e+02 3.173e+02 3.892e+02 5.965e+02, threshold=6.346e+02, percent-clipped=0.0 2022-12-07 14:37:39,652 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:37:40,962 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.98 vs. limit=5.0 2022-12-07 14:38:14,741 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8704, 3.5235, 3.0175, 2.1835, 2.9536, 3.5020, 3.7184, 3.2463], device='cuda:1'), covar=tensor([0.0600, 0.2970, 0.1256, 0.2688, 0.1335, 0.0538, 0.1051, 0.1211], device='cuda:1'), in_proj_covar=tensor([0.0113, 0.0200, 0.0121, 0.0128, 0.0108, 0.0110, 0.0089, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:38:15,099 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 14:38:16,859 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 14:38:25,190 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:38:28,736 INFO [train.py:873] (1/4) Epoch 6, batch 3500, loss[loss=0.1736, simple_loss=0.1895, pruned_loss=0.07887, over 14649.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.1854, pruned_loss=0.07748, over 1887778.31 frames. ], batch size: 23, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:38:43,211 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.629e+02 3.345e+02 4.291e+02 7.270e+02, threshold=6.690e+02, percent-clipped=2.0 2022-12-07 14:38:48,839 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1575, 1.1631, 1.2112, 0.9950, 1.0754, 0.7320, 0.7696, 0.7608], device='cuda:1'), covar=tensor([0.0320, 0.0483, 0.0400, 0.0321, 0.0350, 0.0317, 0.0301, 0.0564], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0011, 0.0015, 0.0012, 0.0016], device='cuda:1'), out_proj_covar=tensor([6.4709e-05, 6.7873e-05, 6.2294e-05, 6.1784e-05, 6.6458e-05, 8.8910e-05, 7.4800e-05, 8.7324e-05], device='cuda:1') 2022-12-07 14:39:06,539 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:08,304 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:14,656 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2022-12-07 14:39:48,946 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:57,988 INFO [train.py:873] (1/4) Epoch 6, batch 3600, loss[loss=0.1849, simple_loss=0.1683, pruned_loss=0.1007, over 2642.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.1849, pruned_loss=0.07617, over 1949105.40 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:40:11,737 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 2.550e+02 3.176e+02 4.214e+02 9.343e+02, threshold=6.353e+02, percent-clipped=2.0 2022-12-07 14:40:30,228 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5663, 2.5034, 4.2836, 4.5718, 4.5879, 2.7930, 4.5020, 3.6443], device='cuda:1'), covar=tensor([0.0111, 0.0434, 0.0424, 0.0133, 0.0105, 0.0646, 0.0112, 0.0395], device='cuda:1'), in_proj_covar=tensor([0.0208, 0.0205, 0.0310, 0.0244, 0.0195, 0.0254, 0.0195, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:40:35,135 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.65 vs. limit=5.0 2022-12-07 14:40:38,976 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:41:05,365 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 14:41:19,898 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 14:41:25,999 INFO [train.py:873] (1/4) Epoch 6, batch 3700, loss[loss=0.2103, simple_loss=0.2078, pruned_loss=0.1063, over 9502.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.1849, pruned_loss=0.0765, over 1891387.96 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:41:34,705 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:41:39,913 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.533e+02 3.248e+02 4.162e+02 7.233e+02, threshold=6.497e+02, percent-clipped=2.0 2022-12-07 14:41:59,610 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8576, 1.5519, 2.0882, 1.7349, 1.9877, 1.5165, 1.6752, 1.8152], device='cuda:1'), covar=tensor([0.1172, 0.2401, 0.0197, 0.1255, 0.0631, 0.1241, 0.0834, 0.0521], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0243, 0.0170, 0.0334, 0.0178, 0.0249, 0.0233, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:42:06,467 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:42:14,700 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4986, 4.3665, 4.8237, 3.9392, 4.5198, 4.8743, 1.5497, 4.2447], device='cuda:1'), covar=tensor([0.0171, 0.0280, 0.0333, 0.0454, 0.0298, 0.0117, 0.3235, 0.0237], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0139, 0.0123, 0.0117, 0.0172, 0.0117, 0.0149, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:42:18,101 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:42:38,990 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7964, 1.6436, 3.0740, 2.2514, 2.8692, 1.6303, 2.3922, 2.7147], device='cuda:1'), covar=tensor([0.0839, 0.4872, 0.0295, 0.5601, 0.0477, 0.3589, 0.1230, 0.0669], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0244, 0.0170, 0.0336, 0.0178, 0.0250, 0.0232, 0.0179], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:42:49,104 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:42:55,067 INFO [train.py:873] (1/4) Epoch 6, batch 3800, loss[loss=0.1837, simple_loss=0.1902, pruned_loss=0.08863, over 6920.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.1853, pruned_loss=0.07704, over 1920194.66 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 16.0 2022-12-07 14:43:09,833 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.547e+02 3.241e+02 4.303e+02 1.041e+03, threshold=6.482e+02, percent-clipped=5.0 2022-12-07 14:43:33,300 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:43:34,376 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:44:15,983 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:44:16,035 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:44:24,886 INFO [train.py:873] (1/4) Epoch 6, batch 3900, loss[loss=0.2213, simple_loss=0.1895, pruned_loss=0.1266, over 1205.00 frames. ], tot_loss[loss=0.168, simple_loss=0.1839, pruned_loss=0.07608, over 1887317.55 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 16.0 2022-12-07 14:44:28,604 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:44:38,880 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.642e+02 3.297e+02 4.041e+02 8.886e+02, threshold=6.594e+02, percent-clipped=2.0 2022-12-07 14:44:58,831 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:06,677 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:48,041 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3904, 2.1014, 2.6564, 1.6452, 1.8325, 2.3043, 1.3021, 2.3729], device='cuda:1'), covar=tensor([0.0727, 0.1332, 0.0744, 0.2042, 0.2534, 0.0791, 0.4781, 0.0811], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0077, 0.0076, 0.0082, 0.0106, 0.0067, 0.0131, 0.0072], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:45:49,541 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:54,441 INFO [train.py:873] (1/4) Epoch 6, batch 4000, loss[loss=0.2007, simple_loss=0.2047, pruned_loss=0.09836, over 9483.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.1839, pruned_loss=0.07624, over 1903856.25 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 16.0 2022-12-07 14:46:08,692 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.207e+02 2.921e+02 3.758e+02 8.130e+02, threshold=5.842e+02, percent-clipped=2.0 2022-12-07 14:46:32,626 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5706, 4.3552, 4.1383, 4.5749, 4.2380, 3.7105, 4.6342, 4.5149], device='cuda:1'), covar=tensor([0.0607, 0.0542, 0.0663, 0.0548, 0.0557, 0.0584, 0.0570, 0.0628], device='cuda:1'), in_proj_covar=tensor([0.0114, 0.0102, 0.0119, 0.0121, 0.0119, 0.0093, 0.0134, 0.0114], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 14:47:18,580 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:47:22,530 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 14:47:22,765 INFO [train.py:873] (1/4) Epoch 6, batch 4100, loss[loss=0.1533, simple_loss=0.1748, pruned_loss=0.06587, over 14248.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.183, pruned_loss=0.07591, over 1890171.81 frames. ], batch size: 80, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:47:37,464 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.509e+02 3.055e+02 4.404e+02 7.346e+02, threshold=6.111e+02, percent-clipped=4.0 2022-12-07 14:47:40,710 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2022-12-07 14:47:55,728 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:48:11,833 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:48:33,015 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2022-12-07 14:48:49,741 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:48:50,444 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:48:51,185 INFO [train.py:873] (1/4) Epoch 6, batch 4200, loss[loss=0.1546, simple_loss=0.1799, pruned_loss=0.06459, over 14216.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.1832, pruned_loss=0.07523, over 1934374.47 frames. ], batch size: 80, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:49:00,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6340, 2.3603, 3.0460, 1.8464, 2.1658, 2.3557, 1.3639, 2.3706], device='cuda:1'), covar=tensor([0.1004, 0.1264, 0.1031, 0.3137, 0.2532, 0.1163, 0.6266, 0.1546], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0076, 0.0077, 0.0082, 0.0106, 0.0067, 0.0131, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:49:07,157 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.446e+02 2.869e+02 3.541e+02 9.255e+02, threshold=5.737e+02, percent-clipped=3.0 2022-12-07 14:49:17,577 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0926, 2.0032, 1.6107, 1.6940, 2.0166, 2.0085, 2.0139, 2.0037], device='cuda:1'), covar=tensor([0.0912, 0.0885, 0.2913, 0.3190, 0.1144, 0.1130, 0.1543, 0.1015], device='cuda:1'), in_proj_covar=tensor([0.0305, 0.0221, 0.0369, 0.0462, 0.0269, 0.0337, 0.0335, 0.0285], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:49:25,507 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9536, 1.3558, 2.7633, 2.5481, 2.6983, 2.7478, 2.0411, 2.7831], device='cuda:1'), covar=tensor([0.0763, 0.1031, 0.0092, 0.0225, 0.0212, 0.0107, 0.0315, 0.0120], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0156, 0.0108, 0.0150, 0.0123, 0.0125, 0.0098, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 14:49:26,861 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 14:49:42,139 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6573, 3.4472, 3.2140, 3.2590, 3.5680, 3.5484, 3.6945, 3.5997], device='cuda:1'), covar=tensor([0.0899, 0.0617, 0.2021, 0.2678, 0.0722, 0.0741, 0.0921, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0305, 0.0221, 0.0368, 0.0461, 0.0267, 0.0336, 0.0334, 0.0283], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:49:42,214 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5031, 2.2191, 2.4089, 1.3870, 2.1156, 2.1838, 2.6094, 2.0638], device='cuda:1'), covar=tensor([0.0621, 0.1465, 0.1028, 0.2366, 0.1019, 0.0683, 0.0422, 0.1780], device='cuda:1'), in_proj_covar=tensor([0.0114, 0.0202, 0.0120, 0.0126, 0.0108, 0.0112, 0.0091, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:50:21,093 INFO [train.py:873] (1/4) Epoch 6, batch 4300, loss[loss=0.1485, simple_loss=0.1831, pruned_loss=0.05694, over 14002.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.184, pruned_loss=0.07573, over 1913457.57 frames. ], batch size: 26, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:50:35,778 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.072e+01 2.450e+02 2.996e+02 3.613e+02 8.545e+02, threshold=5.992e+02, percent-clipped=0.0 2022-12-07 14:50:51,596 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7572, 1.7450, 1.9927, 1.3097, 1.4450, 1.8620, 1.0498, 1.6884], device='cuda:1'), covar=tensor([0.0812, 0.1302, 0.0633, 0.1928, 0.2607, 0.0523, 0.3155, 0.0803], device='cuda:1'), in_proj_covar=tensor([0.0070, 0.0077, 0.0078, 0.0082, 0.0107, 0.0068, 0.0133, 0.0073], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:51:16,346 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2022-12-07 14:51:51,141 INFO [train.py:873] (1/4) Epoch 6, batch 4400, loss[loss=0.159, simple_loss=0.1464, pruned_loss=0.08582, over 1326.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.1841, pruned_loss=0.07554, over 1929243.18 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:52:06,374 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.325e+02 2.637e+02 3.076e+02 3.933e+02 7.884e+02, threshold=6.152e+02, percent-clipped=2.0 2022-12-07 14:52:28,454 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5362, 4.4572, 4.9632, 3.8851, 4.5785, 4.9625, 1.8484, 4.3782], device='cuda:1'), covar=tensor([0.0187, 0.0290, 0.0308, 0.0610, 0.0269, 0.0123, 0.3043, 0.0236], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0142, 0.0125, 0.0119, 0.0175, 0.0121, 0.0152, 0.0162], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:52:30,531 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5391, 2.2601, 2.2632, 1.4426, 2.1627, 2.3196, 2.4814, 2.0554], device='cuda:1'), covar=tensor([0.0687, 0.1529, 0.1057, 0.2276, 0.0997, 0.0635, 0.0701, 0.1720], device='cuda:1'), in_proj_covar=tensor([0.0113, 0.0199, 0.0121, 0.0123, 0.0109, 0.0112, 0.0091, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 14:52:36,335 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:53:14,114 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:53:19,668 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:53:20,405 INFO [train.py:873] (1/4) Epoch 6, batch 4500, loss[loss=0.1541, simple_loss=0.1477, pruned_loss=0.08028, over 2600.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.1837, pruned_loss=0.07524, over 1892943.10 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:53:35,323 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.162e+02 2.450e+02 2.856e+02 3.817e+02 6.882e+02, threshold=5.713e+02, percent-clipped=1.0 2022-12-07 14:53:43,551 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 14:53:57,485 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2653, 2.2527, 3.3459, 3.4508, 3.3694, 2.3050, 3.3860, 2.6119], device='cuda:1'), covar=tensor([0.0145, 0.0324, 0.0269, 0.0159, 0.0122, 0.0496, 0.0108, 0.0372], device='cuda:1'), in_proj_covar=tensor([0.0208, 0.0204, 0.0307, 0.0243, 0.0194, 0.0251, 0.0196, 0.0239], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 14:54:02,622 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:54:22,176 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:54:50,839 INFO [train.py:873] (1/4) Epoch 6, batch 4600, loss[loss=0.1817, simple_loss=0.1905, pruned_loss=0.08645, over 14179.00 frames. ], tot_loss[loss=0.168, simple_loss=0.1844, pruned_loss=0.07581, over 1907776.14 frames. ], batch size: 99, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:55:06,686 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.342e+02 3.312e+02 4.722e+02 1.011e+03, threshold=6.623e+02, percent-clipped=9.0 2022-12-07 14:55:17,235 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:56:15,935 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:56:16,341 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 14:56:20,489 INFO [train.py:873] (1/4) Epoch 6, batch 4700, loss[loss=0.1461, simple_loss=0.1732, pruned_loss=0.05951, over 14311.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.1838, pruned_loss=0.07543, over 1914403.63 frames. ], batch size: 31, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:56:22,843 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 14:56:28,232 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-07 14:56:35,464 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.212e+02 2.979e+02 3.531e+02 6.343e+02, threshold=5.959e+02, percent-clipped=0.0 2022-12-07 14:57:03,458 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2378, 1.2721, 2.6839, 1.2639, 2.5343, 2.5196, 1.9311, 2.5846], device='cuda:1'), covar=tensor([0.0453, 0.3267, 0.0329, 0.2623, 0.0499, 0.0526, 0.1007, 0.0380], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0156, 0.0140, 0.0168, 0.0153, 0.0150, 0.0122, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 14:57:06,055 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:10,566 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:10,868 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=12.10 vs. limit=5.0 2022-12-07 14:57:19,524 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-07 14:57:28,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0247, 1.6280, 3.1497, 2.2178, 3.0549, 1.6275, 2.3851, 2.9207], device='cuda:1'), covar=tensor([0.0919, 0.6494, 0.0557, 0.9066, 0.0535, 0.4891, 0.1716, 0.0444], device='cuda:1'), in_proj_covar=tensor([0.0224, 0.0246, 0.0173, 0.0342, 0.0184, 0.0253, 0.0239, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 14:57:33,846 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 14:57:43,863 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:57:48,833 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:49,699 INFO [train.py:873] (1/4) Epoch 6, batch 4800, loss[loss=0.1477, simple_loss=0.1782, pruned_loss=0.05861, over 14420.00 frames. ], tot_loss[loss=0.167, simple_loss=0.1841, pruned_loss=0.07496, over 2005409.89 frames. ], batch size: 41, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:58:05,121 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.352e+02 2.966e+02 3.514e+02 5.979e+02, threshold=5.933e+02, percent-clipped=1.0 2022-12-07 14:58:23,570 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:58:25,950 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:58:49,637 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6236, 3.0814, 4.1411, 3.1491, 4.3020, 4.2587, 4.1967, 3.7109], device='cuda:1'), covar=tensor([0.0351, 0.2703, 0.0872, 0.1849, 0.0748, 0.0547, 0.1275, 0.1902], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0333, 0.0375, 0.0316, 0.0362, 0.0298, 0.0350, 0.0348], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 14:59:17,983 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:59:18,818 INFO [train.py:873] (1/4) Epoch 6, batch 4900, loss[loss=0.19, simple_loss=0.1871, pruned_loss=0.09647, over 3849.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.1839, pruned_loss=0.07538, over 2006034.26 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:59:33,471 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.456e+02 3.213e+02 4.119e+02 7.897e+02, threshold=6.426e+02, percent-clipped=4.0 2022-12-07 14:59:40,159 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:59:44,603 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.55 vs. limit=5.0 2022-12-07 15:00:47,922 INFO [train.py:873] (1/4) Epoch 6, batch 5000, loss[loss=0.1584, simple_loss=0.1568, pruned_loss=0.07996, over 2674.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.1835, pruned_loss=0.07512, over 1943210.27 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:00:48,073 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7497, 0.7467, 0.6026, 0.7001, 0.8801, 0.5648, 0.7767, 0.8383], device='cuda:1'), covar=tensor([0.0873, 0.0681, 0.0306, 0.0796, 0.0507, 0.0606, 0.0795, 0.0685], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0019, 0.0018, 0.0019, 0.0019, 0.0027, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.3640e-05, 8.5194e-05, 7.8949e-05, 8.2369e-05, 8.4392e-05, 1.0895e-04, 9.0427e-05, 8.2766e-05], device='cuda:1') 2022-12-07 15:01:03,357 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.594e+01 2.343e+02 3.154e+02 4.045e+02 6.991e+02, threshold=6.308e+02, percent-clipped=1.0 2022-12-07 15:01:30,787 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:01:33,648 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:01:47,014 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9246, 4.7166, 5.2155, 4.5177, 4.7917, 5.2368, 1.7806, 4.5928], device='cuda:1'), covar=tensor([0.0235, 0.0270, 0.0272, 0.0432, 0.0333, 0.0120, 0.3233, 0.0233], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0144, 0.0124, 0.0119, 0.0174, 0.0119, 0.0152, 0.0163], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:01:48,796 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:02:16,483 INFO [train.py:873] (1/4) Epoch 6, batch 5100, loss[loss=0.2177, simple_loss=0.179, pruned_loss=0.1282, over 1327.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.1827, pruned_loss=0.07426, over 1976437.67 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:02:24,110 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:02:30,812 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.401e+02 2.989e+02 4.009e+02 7.834e+02, threshold=5.978e+02, percent-clipped=3.0 2022-12-07 15:02:42,273 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:03:03,301 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1724, 3.1225, 3.2142, 3.1284, 3.1618, 2.9776, 1.3652, 3.0073], device='cuda:1'), covar=tensor([0.0377, 0.0456, 0.0658, 0.0445, 0.0475, 0.0647, 0.3513, 0.0393], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0147, 0.0126, 0.0121, 0.0175, 0.0120, 0.0154, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:03:40,340 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:03:45,519 INFO [train.py:873] (1/4) Epoch 6, batch 5200, loss[loss=0.1622, simple_loss=0.1838, pruned_loss=0.07036, over 14616.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.1834, pruned_loss=0.07476, over 2013133.04 frames. ], batch size: 33, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:04:01,018 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.408e+02 3.132e+02 3.708e+02 6.690e+02, threshold=6.264e+02, percent-clipped=2.0 2022-12-07 15:04:03,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5280, 3.2829, 3.2692, 3.5917, 3.1810, 3.0082, 3.5770, 3.4449], device='cuda:1'), covar=tensor([0.0717, 0.0924, 0.0812, 0.0600, 0.1007, 0.0678, 0.0608, 0.0737], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0104, 0.0120, 0.0120, 0.0123, 0.0094, 0.0133, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 15:04:07,530 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:04:07,861 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 15:04:48,395 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 15:04:50,236 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:05:15,649 INFO [train.py:873] (1/4) Epoch 6, batch 5300, loss[loss=0.1245, simple_loss=0.1507, pruned_loss=0.04912, over 13912.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.1831, pruned_loss=0.07401, over 2032932.67 frames. ], batch size: 20, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:05:26,692 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8647, 1.2972, 1.3348, 1.2692, 1.1017, 1.3842, 1.0434, 0.9563], device='cuda:1'), covar=tensor([0.1373, 0.1001, 0.0131, 0.0276, 0.0723, 0.0287, 0.1141, 0.0758], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0063, 0.0051, 0.0056, 0.0079, 0.0059, 0.0085, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:05:29,870 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.504e+02 3.046e+02 3.756e+02 7.408e+02, threshold=6.093e+02, percent-clipped=3.0 2022-12-07 15:06:00,653 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:20,363 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-07 15:06:21,789 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7901, 1.3563, 3.0375, 2.9155, 3.0716, 3.0449, 2.3884, 2.9047], device='cuda:1'), covar=tensor([0.1404, 0.1596, 0.0210, 0.0288, 0.0245, 0.0207, 0.0268, 0.0322], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0155, 0.0108, 0.0149, 0.0123, 0.0125, 0.0097, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 15:06:29,611 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5230, 1.5296, 1.4936, 1.0919, 1.3947, 1.1882, 1.3124, 0.8964], device='cuda:1'), covar=tensor([0.0691, 0.0866, 0.0738, 0.0463, 0.0525, 0.0339, 0.0346, 0.0992], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0011, 0.0015, 0.0012, 0.0016], device='cuda:1'), out_proj_covar=tensor([6.5946e-05, 7.0700e-05, 6.4114e-05, 6.3911e-05, 6.8741e-05, 9.2119e-05, 7.7421e-05, 8.9148e-05], device='cuda:1') 2022-12-07 15:06:43,868 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:44,600 INFO [train.py:873] (1/4) Epoch 6, batch 5400, loss[loss=0.1697, simple_loss=0.1893, pruned_loss=0.07505, over 14264.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.183, pruned_loss=0.07393, over 1999578.64 frames. ], batch size: 57, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:06:47,986 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:59,609 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.808e+01 2.467e+02 2.944e+02 3.777e+02 9.316e+02, threshold=5.888e+02, percent-clipped=3.0 2022-12-07 15:07:06,148 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:07:13,299 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6776, 1.5787, 2.9021, 2.0964, 2.8263, 1.6443, 2.0925, 2.6412], device='cuda:1'), covar=tensor([0.0948, 0.4791, 0.0433, 0.6488, 0.0456, 0.3970, 0.1657, 0.0531], device='cuda:1'), in_proj_covar=tensor([0.0221, 0.0243, 0.0170, 0.0331, 0.0184, 0.0250, 0.0237, 0.0178], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:08:07,813 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2022-12-07 15:08:08,124 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0196, 1.5185, 4.0534, 1.5039, 3.9316, 4.0044, 3.2458, 4.4370], device='cuda:1'), covar=tensor([0.0201, 0.2922, 0.0323, 0.2433, 0.0368, 0.0345, 0.0581, 0.0131], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0155, 0.0137, 0.0169, 0.0152, 0.0151, 0.0121, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:08:08,950 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:08:14,522 INFO [train.py:873] (1/4) Epoch 6, batch 5500, loss[loss=0.1637, simple_loss=0.1732, pruned_loss=0.07704, over 5973.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.182, pruned_loss=0.07329, over 1941472.14 frames. ], batch size: 100, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:08:28,979 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.280e+02 2.901e+02 3.837e+02 7.523e+02, threshold=5.802e+02, percent-clipped=2.0 2022-12-07 15:08:52,292 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:09:43,894 INFO [train.py:873] (1/4) Epoch 6, batch 5600, loss[loss=0.1686, simple_loss=0.1875, pruned_loss=0.07491, over 14335.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.184, pruned_loss=0.07576, over 1901530.11 frames. ], batch size: 73, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:09:55,015 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.57 vs. limit=2.0 2022-12-07 15:09:59,514 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.262e+02 2.794e+02 3.488e+02 5.213e+02, threshold=5.588e+02, percent-clipped=0.0 2022-12-07 15:10:05,188 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4935, 5.2658, 5.1204, 5.5103, 5.2959, 4.6862, 5.4916, 5.4929], device='cuda:1'), covar=tensor([0.0649, 0.0538, 0.0507, 0.0530, 0.0494, 0.0397, 0.0705, 0.0515], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0104, 0.0118, 0.0122, 0.0122, 0.0095, 0.0133, 0.0115], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 15:10:21,698 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1171, 2.1124, 4.0655, 2.8647, 3.9912, 2.0496, 3.1846, 3.9457], device='cuda:1'), covar=tensor([0.0389, 0.4955, 0.0264, 0.7522, 0.0440, 0.3743, 0.1251, 0.0203], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0247, 0.0174, 0.0337, 0.0185, 0.0253, 0.0240, 0.0180], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:10:27,868 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7599, 1.1906, 2.4531, 2.2713, 2.4022, 2.3898, 1.9270, 2.4708], device='cuda:1'), covar=tensor([0.0651, 0.0919, 0.0097, 0.0241, 0.0202, 0.0096, 0.0279, 0.0121], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0154, 0.0108, 0.0148, 0.0123, 0.0123, 0.0097, 0.0102], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 15:10:48,597 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7200, 1.9547, 2.5485, 2.3115, 2.6670, 2.5090, 2.3676, 2.2427], device='cuda:1'), covar=tensor([0.0365, 0.2521, 0.0684, 0.1518, 0.0618, 0.0829, 0.1045, 0.1582], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0328, 0.0370, 0.0308, 0.0354, 0.0299, 0.0350, 0.0336], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:11:13,943 INFO [train.py:873] (1/4) Epoch 6, batch 5700, loss[loss=0.1662, simple_loss=0.1913, pruned_loss=0.07057, over 14263.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.1835, pruned_loss=0.07557, over 1894594.17 frames. ], batch size: 63, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:11:15,302 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 15:11:16,610 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:17,387 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:28,309 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.452e+02 3.294e+02 4.312e+02 1.114e+03, threshold=6.587e+02, percent-clipped=10.0 2022-12-07 15:11:34,604 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:59,856 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:08,970 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 15:12:10,422 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:15,135 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=12.27 vs. limit=5.0 2022-12-07 15:12:17,485 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:18,896 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 15:12:34,533 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3990, 4.0510, 3.9450, 4.4313, 4.1827, 3.6917, 4.4377, 3.7165], device='cuda:1'), covar=tensor([0.0372, 0.0897, 0.0387, 0.0435, 0.0703, 0.0876, 0.0488, 0.0493], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0230, 0.0153, 0.0145, 0.0154, 0.0122, 0.0229, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:12:42,161 INFO [train.py:873] (1/4) Epoch 6, batch 5800, loss[loss=0.168, simple_loss=0.186, pruned_loss=0.07498, over 14266.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.1834, pruned_loss=0.07511, over 1888655.29 frames. ], batch size: 39, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:12:57,493 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.484e+02 3.216e+02 3.939e+02 1.250e+03, threshold=6.432e+02, percent-clipped=4.0 2022-12-07 15:13:03,185 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1945, 1.3691, 3.4546, 1.3741, 3.0998, 3.3257, 2.4261, 3.5321], device='cuda:1'), covar=tensor([0.0198, 0.2563, 0.0194, 0.2088, 0.0822, 0.0287, 0.0727, 0.0157], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0157, 0.0139, 0.0167, 0.0155, 0.0153, 0.0124, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:13:14,745 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7982, 0.8720, 0.9837, 1.0043, 0.9444, 0.6953, 1.0898, 1.0307], device='cuda:1'), covar=tensor([0.1844, 0.1182, 0.0965, 0.1050, 0.1859, 0.0677, 0.1001, 0.0960], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0020, 0.0020, 0.0020, 0.0020, 0.0028, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.8239e-05, 8.7826e-05, 8.4860e-05, 8.6635e-05, 8.9715e-05, 1.1190e-04, 9.3567e-05, 8.8097e-05], device='cuda:1') 2022-12-07 15:14:12,027 INFO [train.py:873] (1/4) Epoch 6, batch 5900, loss[loss=0.1884, simple_loss=0.1967, pruned_loss=0.09008, over 11231.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.1823, pruned_loss=0.07366, over 1935909.86 frames. ], batch size: 100, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:14:12,433 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9291, 1.5455, 3.1383, 1.4610, 3.2479, 3.0838, 2.2835, 3.2397], device='cuda:1'), covar=tensor([0.0193, 0.2055, 0.0236, 0.1772, 0.0188, 0.0309, 0.0679, 0.0153], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0156, 0.0139, 0.0167, 0.0154, 0.0154, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:14:27,080 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.477e+02 2.996e+02 3.769e+02 7.369e+02, threshold=5.992e+02, percent-clipped=1.0 2022-12-07 15:14:40,449 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9459, 3.9837, 4.1529, 3.6606, 3.9493, 4.2054, 1.6131, 3.6705], device='cuda:1'), covar=tensor([0.0223, 0.0240, 0.0352, 0.0386, 0.0297, 0.0219, 0.2964, 0.0282], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0139, 0.0121, 0.0116, 0.0168, 0.0115, 0.0147, 0.0156], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:15:20,325 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:15:36,246 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4564, 1.6537, 4.2017, 1.8001, 4.1450, 4.3984, 3.8790, 4.7893], device='cuda:1'), covar=tensor([0.0162, 0.2886, 0.0316, 0.2430, 0.0274, 0.0259, 0.0358, 0.0117], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0158, 0.0142, 0.0169, 0.0156, 0.0155, 0.0126, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:15:42,280 INFO [train.py:873] (1/4) Epoch 6, batch 6000, loss[loss=0.1487, simple_loss=0.1706, pruned_loss=0.06343, over 13944.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.1824, pruned_loss=0.07385, over 1959141.57 frames. ], batch size: 20, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:15:42,280 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 15:15:47,570 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5380, 2.6453, 3.3619, 2.9925, 3.4436, 3.3762, 3.2335, 2.9433], device='cuda:1'), covar=tensor([0.0351, 0.2578, 0.0776, 0.1389, 0.0715, 0.0718, 0.1296, 0.2024], device='cuda:1'), in_proj_covar=tensor([0.0288, 0.0325, 0.0375, 0.0308, 0.0354, 0.0297, 0.0346, 0.0336], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:15:51,413 INFO [train.py:905] (1/4) Epoch 6, validation: loss=0.1218, simple_loss=0.1652, pruned_loss=0.03921, over 857387.00 frames. 2022-12-07 15:15:51,413 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 15:16:06,715 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.267e+02 3.074e+02 3.851e+02 8.509e+02, threshold=6.148e+02, percent-clipped=4.0 2022-12-07 15:16:24,527 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:16:44,647 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:17:19,399 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:17:20,998 INFO [train.py:873] (1/4) Epoch 6, batch 6100, loss[loss=0.1982, simple_loss=0.1844, pruned_loss=0.106, over 3867.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.1817, pruned_loss=0.0726, over 1974982.32 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 16.0 2022-12-07 15:17:35,709 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.736e+02 3.463e+02 4.201e+02 1.121e+03, threshold=6.926e+02, percent-clipped=3.0 2022-12-07 15:18:03,861 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:18:14,028 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:18:50,019 INFO [train.py:873] (1/4) Epoch 6, batch 6200, loss[loss=0.1806, simple_loss=0.1867, pruned_loss=0.08727, over 9478.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.1824, pruned_loss=0.0735, over 1965299.13 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 16.0 2022-12-07 15:18:54,098 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3695, 1.8165, 3.5120, 2.5286, 3.3762, 1.8966, 2.7383, 3.2156], device='cuda:1'), covar=tensor([0.0511, 0.4625, 0.0231, 0.6498, 0.0439, 0.3862, 0.1213, 0.0392], device='cuda:1'), in_proj_covar=tensor([0.0218, 0.0241, 0.0169, 0.0335, 0.0185, 0.0257, 0.0237, 0.0182], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:18:58,954 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:19:06,617 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.517e+02 3.185e+02 3.672e+02 8.437e+02, threshold=6.370e+02, percent-clipped=2.0 2022-12-07 15:19:25,263 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7687, 3.1887, 4.4680, 3.1966, 4.5078, 4.4420, 4.1752, 3.7976], device='cuda:1'), covar=tensor([0.0396, 0.3305, 0.0794, 0.2102, 0.0929, 0.0753, 0.2163, 0.3170], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0326, 0.0374, 0.0308, 0.0356, 0.0297, 0.0343, 0.0340], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:20:17,801 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8385, 4.7733, 5.1112, 4.0530, 4.7153, 5.0962, 2.0120, 4.4846], device='cuda:1'), covar=tensor([0.0177, 0.0217, 0.0251, 0.0457, 0.0280, 0.0117, 0.2810, 0.0251], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0141, 0.0123, 0.0118, 0.0170, 0.0118, 0.0150, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:20:19,411 INFO [train.py:873] (1/4) Epoch 6, batch 6300, loss[loss=0.1504, simple_loss=0.1733, pruned_loss=0.06378, over 14250.00 frames. ], tot_loss[loss=0.164, simple_loss=0.182, pruned_loss=0.07302, over 2000408.66 frames. ], batch size: 80, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:20:30,466 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6627, 1.8651, 2.6111, 2.8054, 2.6683, 1.9091, 2.7835, 2.2458], device='cuda:1'), covar=tensor([0.0129, 0.0357, 0.0229, 0.0133, 0.0130, 0.0479, 0.0087, 0.0339], device='cuda:1'), in_proj_covar=tensor([0.0216, 0.0211, 0.0317, 0.0251, 0.0201, 0.0256, 0.0206, 0.0246], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 15:20:35,103 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 2.319e+02 2.963e+02 3.987e+02 7.339e+02, threshold=5.926e+02, percent-clipped=1.0 2022-12-07 15:20:48,219 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:21:03,850 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0100, 1.0074, 1.2447, 1.2295, 1.2241, 0.6727, 1.2345, 1.3585], device='cuda:1'), covar=tensor([0.1070, 0.1184, 0.0439, 0.0987, 0.0691, 0.0679, 0.0593, 0.0762], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0020, 0.0019, 0.0020, 0.0028, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.8678e-05, 9.0910e-05, 8.6034e-05, 8.5618e-05, 9.0624e-05, 1.1580e-04, 9.3825e-05, 8.7859e-05], device='cuda:1') 2022-12-07 15:21:11,796 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:21:48,489 INFO [train.py:873] (1/4) Epoch 6, batch 6400, loss[loss=0.162, simple_loss=0.1836, pruned_loss=0.07018, over 13530.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.1816, pruned_loss=0.07189, over 2049668.76 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:21:55,378 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:22:04,827 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.640e+02 3.353e+02 4.203e+02 1.187e+03, threshold=6.705e+02, percent-clipped=10.0 2022-12-07 15:22:35,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-07 15:22:37,041 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:22:50,298 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5231, 3.9489, 3.1987, 4.9321, 4.4113, 4.5115, 3.9576, 3.4296], device='cuda:1'), covar=tensor([0.0731, 0.1285, 0.4704, 0.0546, 0.0888, 0.1993, 0.1151, 0.3716], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0300, 0.0289, 0.0204, 0.0262, 0.0266, 0.0250, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:23:10,408 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1132, 1.5291, 3.8764, 3.6251, 3.7971, 3.8291, 3.3211, 3.9290], device='cuda:1'), covar=tensor([0.1056, 0.1284, 0.0083, 0.0162, 0.0127, 0.0091, 0.0164, 0.0101], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0108, 0.0151, 0.0124, 0.0125, 0.0099, 0.0103], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-07 15:23:17,595 INFO [train.py:873] (1/4) Epoch 6, batch 6500, loss[loss=0.1479, simple_loss=0.1779, pruned_loss=0.05896, over 14609.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.1823, pruned_loss=0.07319, over 1983904.19 frames. ], batch size: 22, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:23:21,311 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:23:33,243 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.290e+02 3.052e+02 3.998e+02 7.186e+02, threshold=6.105e+02, percent-clipped=1.0 2022-12-07 15:23:43,376 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-07 15:23:58,446 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7309, 0.7944, 0.8362, 0.7257, 0.6194, 0.5447, 0.4958, 0.5775], device='cuda:1'), covar=tensor([0.0098, 0.0118, 0.0076, 0.0100, 0.0207, 0.0291, 0.0184, 0.0346], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0016, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([6.9877e-05, 7.5794e-05, 6.6237e-05, 6.7450e-05, 7.1809e-05, 1.0019e-04, 8.5705e-05, 9.5618e-05], device='cuda:1') 2022-12-07 15:24:24,786 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9548, 2.5706, 3.6176, 2.7530, 3.6018, 3.5963, 3.5788, 2.9473], device='cuda:1'), covar=tensor([0.0428, 0.2993, 0.0981, 0.2266, 0.0911, 0.0822, 0.1584, 0.1813], device='cuda:1'), in_proj_covar=tensor([0.0299, 0.0337, 0.0381, 0.0317, 0.0362, 0.0308, 0.0351, 0.0348], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:24:31,323 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 15:24:43,511 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2022-12-07 15:24:46,950 INFO [train.py:873] (1/4) Epoch 6, batch 6600, loss[loss=0.1839, simple_loss=0.1616, pruned_loss=0.1031, over 1254.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.182, pruned_loss=0.07272, over 1996684.66 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:24:48,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7336, 1.5063, 2.0254, 1.5930, 1.9480, 1.3900, 1.6356, 1.6971], device='cuda:1'), covar=tensor([0.1441, 0.2127, 0.0171, 0.1365, 0.0550, 0.1572, 0.0808, 0.0511], device='cuda:1'), in_proj_covar=tensor([0.0217, 0.0245, 0.0168, 0.0333, 0.0187, 0.0257, 0.0237, 0.0184], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:24:58,654 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2022-12-07 15:25:03,287 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.314e+02 2.475e+02 3.026e+02 3.737e+02 6.916e+02, threshold=6.052e+02, percent-clipped=2.0 2022-12-07 15:25:15,315 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:25:25,295 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3510, 2.1726, 4.4198, 2.9692, 4.1846, 1.9162, 3.0570, 4.0539], device='cuda:1'), covar=tensor([0.0416, 0.5010, 0.0320, 0.8224, 0.0336, 0.4361, 0.1480, 0.0327], device='cuda:1'), in_proj_covar=tensor([0.0216, 0.0243, 0.0167, 0.0331, 0.0186, 0.0254, 0.0237, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:25:58,418 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:26:16,141 INFO [train.py:873] (1/4) Epoch 6, batch 6700, loss[loss=0.1648, simple_loss=0.181, pruned_loss=0.07429, over 14183.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.1827, pruned_loss=0.07339, over 2033527.10 frames. ], batch size: 99, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:26:18,204 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1746, 2.5778, 4.0500, 4.2199, 4.4332, 2.4886, 4.3494, 3.2848], device='cuda:1'), covar=tensor([0.0192, 0.0487, 0.0870, 0.0261, 0.0139, 0.0749, 0.0165, 0.0540], device='cuda:1'), in_proj_covar=tensor([0.0218, 0.0210, 0.0319, 0.0251, 0.0201, 0.0255, 0.0209, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 15:26:30,052 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:26:31,879 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.579e+02 3.217e+02 4.461e+02 7.506e+02, threshold=6.434e+02, percent-clipped=5.0 2022-12-07 15:26:32,066 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0019, 0.8151, 0.9053, 1.0350, 1.0815, 0.5739, 0.9602, 1.0168], device='cuda:1'), covar=tensor([0.0752, 0.0761, 0.0376, 0.0402, 0.0443, 0.0657, 0.0717, 0.0497], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0019, 0.0020, 0.0028, 0.0019, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.6413e-05, 8.9286e-05, 8.5744e-05, 8.5999e-05, 8.9293e-05, 1.1415e-04, 9.1033e-05, 8.5718e-05], device='cuda:1') 2022-12-07 15:26:32,948 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1373, 1.0255, 1.2528, 1.2094, 1.1043, 0.5857, 1.1912, 1.5162], device='cuda:1'), covar=tensor([0.1246, 0.0988, 0.1254, 0.0931, 0.1600, 0.0789, 0.1222, 0.1198], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0019, 0.0020, 0.0028, 0.0019, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.6511e-05, 8.9359e-05, 8.5807e-05, 8.6060e-05, 8.9354e-05, 1.1420e-04, 9.1117e-05, 8.5785e-05], device='cuda:1') 2022-12-07 15:26:51,121 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:04,052 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:06,711 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8399, 2.0428, 2.6886, 2.2563, 2.7220, 2.6557, 2.5822, 2.3343], device='cuda:1'), covar=tensor([0.0407, 0.2729, 0.0838, 0.1804, 0.0595, 0.0708, 0.0990, 0.1899], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0330, 0.0375, 0.0312, 0.0358, 0.0304, 0.0347, 0.0344], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:27:18,256 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3269, 0.9528, 1.3206, 1.2076, 1.3207, 0.6560, 1.3669, 1.7740], device='cuda:1'), covar=tensor([0.1212, 0.1154, 0.1925, 0.2642, 0.0980, 0.0628, 0.1126, 0.0524], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0018, 0.0019, 0.0027, 0.0019, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.4486e-05, 8.7399e-05, 8.3575e-05, 8.3565e-05, 8.6671e-05, 1.1119e-04, 8.8713e-05, 8.3744e-05], device='cuda:1') 2022-12-07 15:27:25,059 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:25,705 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9024, 1.4574, 3.0983, 2.9342, 3.1413, 3.1479, 2.6022, 3.1717], device='cuda:1'), covar=tensor([0.1042, 0.1149, 0.0110, 0.0205, 0.0175, 0.0107, 0.0195, 0.0124], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0109, 0.0152, 0.0125, 0.0126, 0.0100, 0.0104], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:27:45,454 INFO [train.py:873] (1/4) Epoch 6, batch 6800, loss[loss=0.1436, simple_loss=0.1691, pruned_loss=0.05909, over 13949.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.1821, pruned_loss=0.07356, over 1970072.74 frames. ], batch size: 23, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:27:45,699 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:47,295 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:49,184 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:28:01,814 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.228e+02 2.960e+02 3.530e+02 6.520e+02, threshold=5.920e+02, percent-clipped=1.0 2022-12-07 15:28:32,303 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:29:15,083 INFO [train.py:873] (1/4) Epoch 6, batch 6900, loss[loss=0.1663, simple_loss=0.1835, pruned_loss=0.07452, over 14281.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.1831, pruned_loss=0.07422, over 2042163.92 frames. ], batch size: 60, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:29:23,813 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8711, 2.6062, 2.6819, 2.8483, 2.7453, 2.7806, 2.9078, 2.4029], device='cuda:1'), covar=tensor([0.0497, 0.1265, 0.0515, 0.0527, 0.0839, 0.0522, 0.0705, 0.0623], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0232, 0.0151, 0.0146, 0.0154, 0.0121, 0.0228, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:29:31,170 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.434e+02 2.991e+02 3.865e+02 6.867e+02, threshold=5.982e+02, percent-clipped=5.0 2022-12-07 15:29:32,914 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 15:30:06,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8449, 3.2947, 4.5272, 3.5927, 4.5216, 4.3863, 4.3239, 4.0549], device='cuda:1'), covar=tensor([0.0343, 0.2823, 0.0844, 0.1791, 0.0756, 0.0713, 0.1853, 0.2059], device='cuda:1'), in_proj_covar=tensor([0.0291, 0.0326, 0.0369, 0.0304, 0.0350, 0.0299, 0.0339, 0.0335], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:30:25,000 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4022, 2.8434, 3.9960, 3.1562, 4.1482, 4.1043, 4.0234, 3.6781], device='cuda:1'), covar=tensor([0.0487, 0.3227, 0.1109, 0.2036, 0.0820, 0.0749, 0.1733, 0.2019], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0327, 0.0371, 0.0307, 0.0350, 0.0299, 0.0340, 0.0337], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:30:43,884 INFO [train.py:873] (1/4) Epoch 6, batch 7000, loss[loss=0.1604, simple_loss=0.1876, pruned_loss=0.06659, over 14197.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.1826, pruned_loss=0.07426, over 1946851.20 frames. ], batch size: 80, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:31:00,497 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.539e+02 3.130e+02 3.727e+02 8.173e+02, threshold=6.260e+02, percent-clipped=3.0 2022-12-07 15:31:47,416 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:31:47,490 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5011, 2.4801, 3.0406, 2.0077, 1.9956, 2.5186, 1.3779, 2.6321], device='cuda:1'), covar=tensor([0.1095, 0.1062, 0.0678, 0.1758, 0.2578, 0.0887, 0.5170, 0.0962], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0080, 0.0078, 0.0083, 0.0106, 0.0067, 0.0130, 0.0074], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:32:07,764 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 15:32:08,041 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:32:12,183 INFO [train.py:873] (1/4) Epoch 6, batch 7100, loss[loss=0.1714, simple_loss=0.1637, pruned_loss=0.08953, over 2671.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.1818, pruned_loss=0.07337, over 1962743.22 frames. ], batch size: 100, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:32:27,516 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.599e+02 3.099e+02 3.894e+02 6.870e+02, threshold=6.198e+02, percent-clipped=1.0 2022-12-07 15:32:49,207 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8840, 2.6918, 2.0533, 2.9472, 2.7500, 2.8116, 2.4628, 2.1848], device='cuda:1'), covar=tensor([0.0501, 0.1200, 0.3459, 0.0510, 0.0806, 0.0687, 0.1215, 0.2929], device='cuda:1'), in_proj_covar=tensor([0.0237, 0.0304, 0.0291, 0.0207, 0.0268, 0.0271, 0.0255, 0.0283], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:33:05,884 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.40 vs. limit=5.0 2022-12-07 15:33:44,152 INFO [train.py:873] (1/4) Epoch 6, batch 7200, loss[loss=0.1505, simple_loss=0.1783, pruned_loss=0.06135, over 14259.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.1821, pruned_loss=0.07445, over 1907165.39 frames. ], batch size: 80, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:33:48,699 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9857, 2.8567, 2.5907, 2.7379, 2.9080, 2.8932, 2.9861, 2.9801], device='cuda:1'), covar=tensor([0.0877, 0.0673, 0.2203, 0.2669, 0.0903, 0.0911, 0.1253, 0.0873], device='cuda:1'), in_proj_covar=tensor([0.0308, 0.0229, 0.0383, 0.0480, 0.0277, 0.0358, 0.0349, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:33:49,629 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7228, 2.0920, 2.0110, 2.1783, 1.7004, 2.2047, 1.8704, 1.0297], device='cuda:1'), covar=tensor([0.1962, 0.0886, 0.1087, 0.0442, 0.1124, 0.0443, 0.1294, 0.3386], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0061, 0.0049, 0.0052, 0.0078, 0.0055, 0.0081, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:34:00,688 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.403e+02 2.886e+02 3.979e+02 7.470e+02, threshold=5.772e+02, percent-clipped=2.0 2022-12-07 15:34:24,865 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:34:46,881 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:34:51,022 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7271, 3.4189, 2.9430, 1.9650, 3.0983, 3.1938, 3.5624, 2.6968], device='cuda:1'), covar=tensor([0.0611, 0.2368, 0.1405, 0.2972, 0.1021, 0.0680, 0.1313, 0.1875], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0200, 0.0128, 0.0130, 0.0114, 0.0115, 0.0095, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:34:51,909 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1816, 1.1423, 1.3219, 1.0287, 1.1375, 0.9479, 1.0877, 0.8582], device='cuda:1'), covar=tensor([0.0354, 0.0560, 0.0592, 0.0456, 0.0316, 0.0357, 0.0271, 0.0586], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0011, 0.0016, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([7.2418e-05, 7.7148e-05, 6.9609e-05, 7.0853e-05, 7.1680e-05, 1.0159e-04, 8.5679e-05, 9.7270e-05], device='cuda:1') 2022-12-07 15:35:13,053 INFO [train.py:873] (1/4) Epoch 6, batch 7300, loss[loss=0.169, simple_loss=0.1616, pruned_loss=0.0882, over 2658.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.1818, pruned_loss=0.0743, over 1833028.39 frames. ], batch size: 100, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:35:18,620 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 15:35:29,219 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.544e+02 3.092e+02 3.955e+02 6.821e+02, threshold=6.183e+02, percent-clipped=3.0 2022-12-07 15:35:31,977 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1874, 2.0021, 2.5714, 1.7100, 1.7735, 2.2743, 1.2957, 2.2657], device='cuda:1'), covar=tensor([0.1011, 0.1569, 0.0628, 0.2908, 0.2818, 0.0786, 0.4821, 0.0987], device='cuda:1'), in_proj_covar=tensor([0.0072, 0.0082, 0.0079, 0.0086, 0.0108, 0.0068, 0.0134, 0.0076], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:35:41,309 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:17,658 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:37,790 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:41,983 INFO [train.py:873] (1/4) Epoch 6, batch 7400, loss[loss=0.1813, simple_loss=0.2017, pruned_loss=0.08043, over 14466.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.1822, pruned_loss=0.07425, over 1896761.00 frames. ], batch size: 51, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:36:42,984 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8362, 1.4942, 2.0461, 1.7189, 2.0097, 1.4569, 1.7556, 1.8568], device='cuda:1'), covar=tensor([0.1298, 0.2128, 0.0169, 0.1628, 0.0438, 0.1046, 0.0800, 0.0459], device='cuda:1'), in_proj_covar=tensor([0.0217, 0.0242, 0.0170, 0.0328, 0.0185, 0.0246, 0.0234, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:36:52,562 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.5622, 0.6361, 0.6232, 0.6953, 0.6326, 0.3201, 0.6526, 0.7782], device='cuda:1'), covar=tensor([0.0197, 0.0297, 0.0195, 0.0188, 0.0067, 0.0139, 0.0578, 0.0226], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0020, 0.0020, 0.0028, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.8129e-05, 8.9738e-05, 8.5895e-05, 8.8997e-05, 8.9959e-05, 1.1570e-04, 9.3177e-05, 8.9085e-05], device='cuda:1') 2022-12-07 15:36:58,363 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.594e+02 3.335e+02 4.109e+02 6.658e+02, threshold=6.669e+02, percent-clipped=6.0 2022-12-07 15:37:00,158 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:37:20,785 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:37:24,115 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 15:37:43,422 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4751, 2.6748, 3.4905, 2.3838, 2.3188, 2.8364, 1.5727, 2.7487], device='cuda:1'), covar=tensor([0.1422, 0.0892, 0.0552, 0.1518, 0.2138, 0.0941, 0.4755, 0.1615], device='cuda:1'), in_proj_covar=tensor([0.0071, 0.0083, 0.0078, 0.0086, 0.0109, 0.0069, 0.0135, 0.0077], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:38:12,292 INFO [train.py:873] (1/4) Epoch 6, batch 7500, loss[loss=0.1755, simple_loss=0.1959, pruned_loss=0.07753, over 14294.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.182, pruned_loss=0.0735, over 1946415.85 frames. ], batch size: 76, lr: 1.25e-02, grad_scale: 8.0 2022-12-07 15:38:27,608 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.117e+02 2.437e+02 3.096e+02 3.781e+02 7.631e+02, threshold=6.193e+02, percent-clipped=3.0 2022-12-07 15:38:44,195 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 15:38:49,353 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0130, 1.9105, 1.9959, 2.0654, 1.9745, 1.6980, 1.2082, 1.7457], device='cuda:1'), covar=tensor([0.0451, 0.0422, 0.0511, 0.0291, 0.0362, 0.1081, 0.1945, 0.0418], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0149, 0.0128, 0.0120, 0.0175, 0.0119, 0.0154, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:39:51,271 INFO [train.py:873] (1/4) Epoch 7, batch 0, loss[loss=0.1888, simple_loss=0.1969, pruned_loss=0.09035, over 6939.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.1969, pruned_loss=0.09035, over 6939.00 frames. ], batch size: 100, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:39:51,271 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 15:39:59,859 INFO [train.py:905] (1/4) Epoch 7, validation: loss=0.1305, simple_loss=0.175, pruned_loss=0.04304, over 857387.00 frames. 2022-12-07 15:39:59,859 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 15:40:33,641 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 15:40:36,192 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 15:40:51,472 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.719e+01 2.285e+02 3.118e+02 4.102e+02 1.207e+03, threshold=6.237e+02, percent-clipped=4.0 2022-12-07 15:40:59,125 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:41:07,402 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7558, 4.7851, 4.8326, 4.3686, 4.5137, 5.1968, 1.8293, 4.2654], device='cuda:1'), covar=tensor([0.0277, 0.0282, 0.0498, 0.0402, 0.0453, 0.0117, 0.3961, 0.0399], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0148, 0.0127, 0.0119, 0.0174, 0.0120, 0.0153, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:41:27,557 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7609, 2.4238, 4.6889, 3.0604, 4.4709, 2.0203, 3.4231, 4.3443], device='cuda:1'), covar=tensor([0.0382, 0.5039, 0.0273, 0.9596, 0.0362, 0.4298, 0.1457, 0.0270], device='cuda:1'), in_proj_covar=tensor([0.0216, 0.0240, 0.0170, 0.0325, 0.0187, 0.0246, 0.0236, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 15:41:31,134 INFO [train.py:873] (1/4) Epoch 7, batch 100, loss[loss=0.1492, simple_loss=0.1788, pruned_loss=0.0598, over 14283.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.1806, pruned_loss=0.07019, over 900603.17 frames. ], batch size: 66, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:41:58,273 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 15:42:04,739 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 15:42:21,345 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.542e+02 3.020e+02 3.979e+02 7.369e+02, threshold=6.041e+02, percent-clipped=5.0 2022-12-07 15:42:40,325 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7546, 1.3776, 3.6039, 3.4081, 3.5359, 3.6623, 2.9567, 3.5995], device='cuda:1'), covar=tensor([0.1436, 0.1536, 0.0127, 0.0242, 0.0204, 0.0133, 0.0279, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0158, 0.0110, 0.0155, 0.0126, 0.0128, 0.0101, 0.0107], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:43:00,540 INFO [train.py:873] (1/4) Epoch 7, batch 200, loss[loss=0.1562, simple_loss=0.1825, pruned_loss=0.06494, over 14138.00 frames. ], tot_loss[loss=0.162, simple_loss=0.1814, pruned_loss=0.07134, over 1386514.01 frames. ], batch size: 84, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:43:44,787 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7129, 3.6806, 3.2228, 3.3556, 2.6766, 3.9242, 3.2651, 1.6964], device='cuda:1'), covar=tensor([0.2506, 0.0527, 0.1994, 0.1669, 0.1192, 0.0363, 0.1423, 0.3284], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0061, 0.0051, 0.0054, 0.0079, 0.0057, 0.0083, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:43:50,945 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.408e+02 2.969e+02 3.747e+02 6.560e+02, threshold=5.938e+02, percent-clipped=2.0 2022-12-07 15:44:09,589 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0856, 1.4218, 1.3349, 1.2901, 1.1774, 1.3329, 1.1959, 1.1342], device='cuda:1'), covar=tensor([0.1105, 0.0921, 0.1512, 0.0945, 0.1582, 0.0572, 0.0645, 0.1171], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0013, 0.0011, 0.0011, 0.0011, 0.0016, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([7.2095e-05, 7.8721e-05, 7.1197e-05, 7.1313e-05, 7.2619e-05, 1.0366e-04, 8.7461e-05, 9.9878e-05], device='cuda:1') 2022-12-07 15:44:28,387 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8332, 3.7964, 4.0474, 3.4590, 3.8494, 3.9439, 1.3272, 3.6311], device='cuda:1'), covar=tensor([0.0260, 0.0317, 0.0374, 0.0481, 0.0299, 0.0291, 0.3482, 0.0294], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0151, 0.0132, 0.0123, 0.0178, 0.0122, 0.0155, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:44:30,683 INFO [train.py:873] (1/4) Epoch 7, batch 300, loss[loss=0.1771, simple_loss=0.1909, pruned_loss=0.08165, over 14225.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.1793, pruned_loss=0.07013, over 1658572.30 frames. ], batch size: 94, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:44:34,319 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-07 15:44:56,172 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:44:58,081 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:45:05,223 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:45:14,236 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1331, 1.3478, 1.4348, 1.3920, 1.5772, 1.2645, 1.3459, 1.2409], device='cuda:1'), covar=tensor([0.0977, 0.2036, 0.1025, 0.1087, 0.0729, 0.0465, 0.0389, 0.1216], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0013, 0.0011, 0.0011, 0.0011, 0.0017, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([7.1644e-05, 7.8816e-05, 7.0587e-05, 7.1343e-05, 7.1973e-05, 1.0539e-04, 8.7173e-05, 9.9802e-05], device='cuda:1') 2022-12-07 15:45:20,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.117e+02 2.543e+02 3.312e+02 3.999e+02 6.909e+02, threshold=6.625e+02, percent-clipped=2.0 2022-12-07 15:45:27,754 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:48,344 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:50,990 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:59,856 INFO [train.py:873] (1/4) Epoch 7, batch 400, loss[loss=0.171, simple_loss=0.1739, pruned_loss=0.08406, over 3876.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.18, pruned_loss=0.07107, over 1731281.99 frames. ], batch size: 100, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:46:01,012 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:46:10,766 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:46:17,704 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-07 15:46:42,905 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.36 vs. limit=2.0 2022-12-07 15:46:50,562 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.431e+02 3.067e+02 3.597e+02 6.934e+02, threshold=6.134e+02, percent-clipped=1.0 2022-12-07 15:46:56,413 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:47:07,730 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 15:47:17,585 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:47:30,182 INFO [train.py:873] (1/4) Epoch 7, batch 500, loss[loss=0.2042, simple_loss=0.204, pruned_loss=0.1022, over 9428.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.1809, pruned_loss=0.07189, over 1851254.84 frames. ], batch size: 100, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:47:37,307 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.65 vs. limit=2.0 2022-12-07 15:47:47,255 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 15:48:12,854 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:48:21,753 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.455e+02 3.322e+02 4.008e+02 8.135e+02, threshold=6.645e+02, percent-clipped=7.0 2022-12-07 15:48:35,297 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 15:48:59,908 INFO [train.py:873] (1/4) Epoch 7, batch 600, loss[loss=0.1944, simple_loss=0.1993, pruned_loss=0.09476, over 4987.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.1804, pruned_loss=0.07191, over 1872852.12 frames. ], batch size: 100, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:49:27,331 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:49:31,409 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.58 vs. limit=5.0 2022-12-07 15:49:42,065 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6045, 4.4659, 4.7949, 3.9212, 4.5615, 4.9187, 1.5529, 4.2933], device='cuda:1'), covar=tensor([0.0172, 0.0236, 0.0315, 0.0391, 0.0237, 0.0098, 0.3177, 0.0234], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0147, 0.0127, 0.0121, 0.0175, 0.0120, 0.0152, 0.0162], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:49:49,709 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.803e+01 2.347e+02 3.029e+02 4.189e+02 7.825e+02, threshold=6.059e+02, percent-clipped=1.0 2022-12-07 15:50:01,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7764, 1.3976, 1.7805, 1.3179, 1.4234, 1.8285, 1.5339, 1.5669], device='cuda:1'), covar=tensor([0.0454, 0.0878, 0.0522, 0.1038, 0.0887, 0.0604, 0.0472, 0.1441], device='cuda:1'), in_proj_covar=tensor([0.0114, 0.0191, 0.0122, 0.0125, 0.0110, 0.0111, 0.0093, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:50:09,874 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:50:15,097 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:50:28,495 INFO [train.py:873] (1/4) Epoch 7, batch 700, loss[loss=0.1663, simple_loss=0.185, pruned_loss=0.07382, over 14266.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.1804, pruned_loss=0.0723, over 1917740.30 frames. ], batch size: 80, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:51:18,727 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.605e+01 2.609e+02 3.339e+02 4.151e+02 6.051e+02, threshold=6.678e+02, percent-clipped=0.0 2022-12-07 15:51:18,841 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:51:51,529 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2022-12-07 15:51:56,447 INFO [train.py:873] (1/4) Epoch 7, batch 800, loss[loss=0.1511, simple_loss=0.1694, pruned_loss=0.06639, over 13478.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.1808, pruned_loss=0.07265, over 1900539.01 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:52:34,717 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:52:47,697 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.967e+01 2.522e+02 3.122e+02 4.007e+02 6.928e+02, threshold=6.244e+02, percent-clipped=1.0 2022-12-07 15:53:26,506 INFO [train.py:873] (1/4) Epoch 7, batch 900, loss[loss=0.1567, simple_loss=0.1828, pruned_loss=0.06533, over 14428.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.1797, pruned_loss=0.0715, over 1955579.07 frames. ], batch size: 41, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:53:29,271 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:53:46,810 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8450, 1.3935, 1.7644, 1.4295, 1.3504, 1.8999, 1.6179, 1.6036], device='cuda:1'), covar=tensor([0.0583, 0.1236, 0.0610, 0.0929, 0.1645, 0.0736, 0.0550, 0.1689], device='cuda:1'), in_proj_covar=tensor([0.0117, 0.0195, 0.0124, 0.0126, 0.0114, 0.0114, 0.0095, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:54:16,875 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.538e+02 3.268e+02 4.038e+02 9.582e+02, threshold=6.536e+02, percent-clipped=3.0 2022-12-07 15:54:23,205 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:54:42,249 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:54:55,205 INFO [train.py:873] (1/4) Epoch 7, batch 1000, loss[loss=0.2689, simple_loss=0.2105, pruned_loss=0.1637, over 1217.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.1803, pruned_loss=0.07245, over 1905724.31 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:55:00,587 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6234, 4.5694, 4.8176, 4.1698, 4.6458, 4.9317, 1.5674, 4.4367], device='cuda:1'), covar=tensor([0.0183, 0.0227, 0.0387, 0.0337, 0.0254, 0.0128, 0.3217, 0.0212], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0145, 0.0126, 0.0119, 0.0173, 0.0120, 0.0150, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:55:24,798 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:55:46,198 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.372e+02 2.921e+02 3.865e+02 1.419e+03, threshold=5.842e+02, percent-clipped=9.0 2022-12-07 15:55:46,343 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:56:11,378 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2705, 1.6057, 4.7013, 4.3194, 4.1410, 4.6762, 4.3305, 4.7650], device='cuda:1'), covar=tensor([0.1073, 0.1334, 0.0063, 0.0147, 0.0153, 0.0084, 0.0105, 0.0073], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0157, 0.0111, 0.0155, 0.0126, 0.0127, 0.0102, 0.0105], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:56:14,666 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7305, 0.5579, 0.5346, 0.7010, 0.5292, 0.2642, 0.4417, 0.6064], device='cuda:1'), covar=tensor([0.0135, 0.0142, 0.0114, 0.0123, 0.0368, 0.0376, 0.0237, 0.0497], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0010, 0.0012, 0.0016, 0.0013, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.2401e-05, 7.8550e-05, 7.1329e-05, 7.1174e-05, 7.3794e-05, 1.0401e-04, 8.8507e-05, 1.0197e-04], device='cuda:1') 2022-12-07 15:56:18,080 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1838, 2.5787, 4.1211, 4.2152, 4.4274, 2.6687, 4.3309, 3.4679], device='cuda:1'), covar=tensor([0.0132, 0.0381, 0.0542, 0.0179, 0.0103, 0.0614, 0.0115, 0.0386], device='cuda:1'), in_proj_covar=tensor([0.0220, 0.0210, 0.0319, 0.0255, 0.0203, 0.0257, 0.0214, 0.0248], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 15:56:24,648 INFO [train.py:873] (1/4) Epoch 7, batch 1100, loss[loss=0.1773, simple_loss=0.1642, pruned_loss=0.09516, over 2602.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.1801, pruned_loss=0.07218, over 1898639.51 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:56:29,019 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:56:29,827 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6325, 3.4103, 3.2171, 3.2364, 3.5574, 3.5283, 3.6470, 3.5682], device='cuda:1'), covar=tensor([0.0867, 0.0632, 0.2002, 0.2906, 0.0735, 0.0794, 0.0967, 0.0952], device='cuda:1'), in_proj_covar=tensor([0.0322, 0.0233, 0.0393, 0.0495, 0.0283, 0.0359, 0.0346, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 15:56:44,615 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2022-12-07 15:57:01,960 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:57:12,093 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5348, 3.0941, 2.4095, 3.6951, 3.4209, 3.4691, 2.9500, 2.3382], device='cuda:1'), covar=tensor([0.0805, 0.1583, 0.4306, 0.0335, 0.0888, 0.1012, 0.1441, 0.4758], device='cuda:1'), in_proj_covar=tensor([0.0236, 0.0297, 0.0282, 0.0201, 0.0262, 0.0264, 0.0251, 0.0272], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 15:57:15,197 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.418e+02 3.079e+02 3.635e+02 7.997e+02, threshold=6.158e+02, percent-clipped=1.0 2022-12-07 15:57:33,912 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9915, 5.3913, 5.4470, 5.9164, 5.4913, 4.6893, 5.9005, 4.8987], device='cuda:1'), covar=tensor([0.0295, 0.0738, 0.0240, 0.0370, 0.0544, 0.0343, 0.0340, 0.0414], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0234, 0.0153, 0.0146, 0.0153, 0.0124, 0.0228, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 15:57:36,108 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0128, 3.5668, 3.1297, 2.3698, 3.2055, 3.4663, 3.8329, 3.1890], device='cuda:1'), covar=tensor([0.0536, 0.2892, 0.1353, 0.2649, 0.1219, 0.0646, 0.1294, 0.1643], device='cuda:1'), in_proj_covar=tensor([0.0112, 0.0192, 0.0121, 0.0125, 0.0110, 0.0112, 0.0094, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-07 15:57:45,029 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:57:53,829 INFO [train.py:873] (1/4) Epoch 7, batch 1200, loss[loss=0.1438, simple_loss=0.168, pruned_loss=0.05978, over 13868.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.1798, pruned_loss=0.07098, over 2014876.46 frames. ], batch size: 20, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:58:13,385 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:58:20,040 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:58:41,451 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 15:58:42,659 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 15:58:44,673 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.081e+01 2.469e+02 2.984e+02 3.753e+02 8.629e+02, threshold=5.968e+02, percent-clipped=3.0 2022-12-07 15:58:46,460 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:58:48,018 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 15:59:08,361 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 15:59:14,287 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:59:23,323 INFO [train.py:873] (1/4) Epoch 7, batch 1300, loss[loss=0.1569, simple_loss=0.1788, pruned_loss=0.06751, over 14173.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.1792, pruned_loss=0.0713, over 1952897.37 frames. ], batch size: 84, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 16:00:14,512 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.325e+02 3.068e+02 3.769e+02 6.043e+02, threshold=6.136e+02, percent-clipped=2.0 2022-12-07 16:00:53,015 INFO [train.py:873] (1/4) Epoch 7, batch 1400, loss[loss=0.1727, simple_loss=0.1905, pruned_loss=0.07746, over 14593.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.1804, pruned_loss=0.07151, over 1979352.68 frames. ], batch size: 34, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 16:01:27,202 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:01:45,248 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.517e+02 3.054e+02 3.788e+02 7.041e+02, threshold=6.108e+02, percent-clipped=3.0 2022-12-07 16:02:21,992 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:02:23,394 INFO [train.py:873] (1/4) Epoch 7, batch 1500, loss[loss=0.1702, simple_loss=0.1789, pruned_loss=0.08078, over 10320.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.1793, pruned_loss=0.07072, over 2017207.35 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 16:02:40,533 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0784, 1.9562, 1.6986, 1.9488, 1.8250, 2.0315, 1.8819, 1.8927], device='cuda:1'), covar=tensor([0.0610, 0.0857, 0.2070, 0.0471, 0.1020, 0.0404, 0.1501, 0.0686], device='cuda:1'), in_proj_covar=tensor([0.0233, 0.0297, 0.0282, 0.0205, 0.0261, 0.0263, 0.0249, 0.0268], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:02:52,153 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.35 vs. limit=2.0 2022-12-07 16:02:57,695 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9500, 2.8243, 2.0917, 3.0323, 2.7445, 2.8742, 2.5535, 2.2940], device='cuda:1'), covar=tensor([0.0705, 0.1179, 0.3505, 0.0379, 0.0918, 0.0785, 0.1184, 0.2774], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0296, 0.0280, 0.0205, 0.0261, 0.0262, 0.0248, 0.0269], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:03:13,535 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.064e+01 2.462e+02 3.053e+02 4.274e+02 1.237e+03, threshold=6.105e+02, percent-clipped=6.0 2022-12-07 16:03:15,395 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:03:32,160 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 16:03:38,717 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:03:46,587 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2461, 1.8924, 2.3298, 2.4202, 2.1116, 1.8656, 2.4964, 2.0710], device='cuda:1'), covar=tensor([0.0146, 0.0308, 0.0151, 0.0115, 0.0183, 0.0399, 0.0119, 0.0228], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0214, 0.0323, 0.0258, 0.0204, 0.0259, 0.0217, 0.0250], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:03:46,832 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-07 16:03:51,575 INFO [train.py:873] (1/4) Epoch 7, batch 1600, loss[loss=0.1577, simple_loss=0.1557, pruned_loss=0.07985, over 3865.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.1792, pruned_loss=0.07148, over 1939471.85 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:03:57,830 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:04:42,980 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 2.534e+02 3.085e+02 3.879e+02 6.356e+02, threshold=6.171e+02, percent-clipped=1.0 2022-12-07 16:05:21,095 INFO [train.py:873] (1/4) Epoch 7, batch 1700, loss[loss=0.1326, simple_loss=0.155, pruned_loss=0.0551, over 13664.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1784, pruned_loss=0.07033, over 1934416.09 frames. ], batch size: 17, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:05:48,099 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6441, 2.0508, 2.5530, 2.3251, 2.6798, 2.5007, 2.4729, 2.3407], device='cuda:1'), covar=tensor([0.0553, 0.2595, 0.0732, 0.1874, 0.0521, 0.0688, 0.0964, 0.2041], device='cuda:1'), in_proj_covar=tensor([0.0304, 0.0328, 0.0386, 0.0311, 0.0360, 0.0304, 0.0357, 0.0335], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:06:11,978 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.426e+02 3.035e+02 3.683e+02 7.018e+02, threshold=6.070e+02, percent-clipped=2.0 2022-12-07 16:06:40,295 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.09 vs. limit=2.0 2022-12-07 16:06:44,210 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:06:50,174 INFO [train.py:873] (1/4) Epoch 7, batch 1800, loss[loss=0.1999, simple_loss=0.202, pruned_loss=0.09893, over 10366.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.1785, pruned_loss=0.07013, over 1982349.99 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:07:09,867 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 16:07:23,195 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6015, 5.3485, 5.0411, 5.6413, 5.2184, 4.5067, 5.6638, 5.4797], device='cuda:1'), covar=tensor([0.0655, 0.0786, 0.0743, 0.0581, 0.0688, 0.0491, 0.0654, 0.0712], device='cuda:1'), in_proj_covar=tensor([0.0114, 0.0102, 0.0118, 0.0121, 0.0120, 0.0093, 0.0131, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 16:07:40,890 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.187e+02 2.439e+02 3.276e+02 4.235e+02 9.028e+02, threshold=6.551e+02, percent-clipped=6.0 2022-12-07 16:07:41,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6619, 1.5707, 1.7728, 2.0402, 1.4237, 1.6757, 1.8844, 1.8196], device='cuda:1'), covar=tensor([0.0090, 0.0134, 0.0065, 0.0042, 0.0138, 0.0186, 0.0090, 0.0061], device='cuda:1'), in_proj_covar=tensor([0.0224, 0.0214, 0.0324, 0.0258, 0.0206, 0.0260, 0.0218, 0.0249], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:07:50,921 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:07:59,611 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:05,752 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:19,478 INFO [train.py:873] (1/4) Epoch 7, batch 1900, loss[loss=0.144, simple_loss=0.1733, pruned_loss=0.05736, over 14675.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.1783, pruned_loss=0.07022, over 1989630.37 frames. ], batch size: 33, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:08:23,862 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4805, 4.2271, 3.9810, 4.0789, 4.2666, 4.3759, 4.4720, 4.4499], device='cuda:1'), covar=tensor([0.0816, 0.0601, 0.1826, 0.2679, 0.0694, 0.0615, 0.0851, 0.0808], device='cuda:1'), in_proj_covar=tensor([0.0320, 0.0231, 0.0389, 0.0492, 0.0287, 0.0362, 0.0339, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:08:42,247 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:44,928 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:47,787 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2022-12-07 16:08:48,297 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:09:05,137 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-07 16:09:10,048 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.352e+02 2.909e+02 3.754e+02 8.188e+02, threshold=5.818e+02, percent-clipped=1.0 2022-12-07 16:09:21,420 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1413, 2.8368, 2.8023, 1.8552, 2.5884, 2.7628, 3.1134, 2.5258], device='cuda:1'), covar=tensor([0.0809, 0.2167, 0.1282, 0.2744, 0.1439, 0.0726, 0.0850, 0.2052], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0197, 0.0125, 0.0128, 0.0113, 0.0119, 0.0098, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0007, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 16:09:43,351 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-07 16:09:47,746 INFO [train.py:873] (1/4) Epoch 7, batch 2000, loss[loss=0.1525, simple_loss=0.1751, pruned_loss=0.06495, over 14238.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.1792, pruned_loss=0.07056, over 2009343.05 frames. ], batch size: 89, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:09:53,056 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:10:27,369 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4155, 2.3089, 3.3350, 2.5511, 3.3833, 3.0932, 3.1355, 2.9291], device='cuda:1'), covar=tensor([0.0531, 0.3210, 0.0962, 0.1969, 0.0759, 0.0939, 0.1370, 0.2309], device='cuda:1'), in_proj_covar=tensor([0.0304, 0.0329, 0.0384, 0.0307, 0.0356, 0.0305, 0.0353, 0.0334], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:10:40,009 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.312e+02 2.798e+02 3.545e+02 9.094e+02, threshold=5.596e+02, percent-clipped=2.0 2022-12-07 16:10:48,750 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:11:11,232 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:11:17,999 INFO [train.py:873] (1/4) Epoch 7, batch 2100, loss[loss=0.1441, simple_loss=0.1675, pruned_loss=0.06033, over 14188.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1787, pruned_loss=0.07013, over 1995210.37 frames. ], batch size: 99, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:11:30,037 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-07 16:11:53,876 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:12:09,090 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.308e+02 2.873e+02 3.601e+02 7.245e+02, threshold=5.745e+02, percent-clipped=7.0 2022-12-07 16:12:25,960 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1827, 1.0315, 1.0184, 0.9604, 1.3126, 0.6519, 1.0351, 1.2613], device='cuda:1'), covar=tensor([0.0917, 0.0928, 0.0750, 0.0944, 0.1154, 0.0721, 0.0918, 0.0969], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0020, 0.0019, 0.0028, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.8774e-05, 9.0493e-05, 8.6426e-05, 9.0853e-05, 8.9391e-05, 1.1593e-04, 9.5039e-05, 9.1621e-05], device='cuda:1') 2022-12-07 16:12:43,092 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9117, 1.8141, 4.6679, 2.1176, 4.3780, 4.6925, 4.5888, 5.2833], device='cuda:1'), covar=tensor([0.0187, 0.2616, 0.0335, 0.1961, 0.0294, 0.0321, 0.0235, 0.0127], device='cuda:1'), in_proj_covar=tensor([0.0162, 0.0159, 0.0147, 0.0165, 0.0159, 0.0159, 0.0129, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:12:46,255 INFO [train.py:873] (1/4) Epoch 7, batch 2200, loss[loss=0.1983, simple_loss=0.2098, pruned_loss=0.09344, over 10395.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.1794, pruned_loss=0.07108, over 1981232.32 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:13:08,057 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:13:30,317 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 16:13:37,949 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.347e+02 2.892e+02 3.477e+02 7.491e+02, threshold=5.783e+02, percent-clipped=4.0 2022-12-07 16:13:42,267 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5645, 1.2225, 3.1961, 3.0138, 3.2578, 3.3177, 2.4393, 3.2458], device='cuda:1'), covar=tensor([0.1990, 0.2226, 0.0244, 0.0422, 0.0300, 0.0234, 0.0494, 0.0273], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0155, 0.0111, 0.0153, 0.0125, 0.0127, 0.0101, 0.0107], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 16:13:47,042 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 16:13:55,971 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1038, 1.0099, 0.9796, 0.9439, 1.1071, 0.6541, 0.8835, 1.3508], device='cuda:1'), covar=tensor([0.0948, 0.1465, 0.0440, 0.1261, 0.1429, 0.0732, 0.1411, 0.0809], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0020, 0.0019, 0.0028, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.9152e-05, 9.1309e-05, 8.7750e-05, 9.1471e-05, 9.0683e-05, 1.1744e-04, 9.6596e-05, 9.1865e-05], device='cuda:1') 2022-12-07 16:13:56,829 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8371, 1.5311, 2.1058, 1.7214, 1.9992, 1.4692, 1.7270, 1.7981], device='cuda:1'), covar=tensor([0.1086, 0.2360, 0.0168, 0.1255, 0.0460, 0.1093, 0.0676, 0.0323], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0240, 0.0166, 0.0323, 0.0191, 0.0247, 0.0226, 0.0185], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:14:15,734 INFO [train.py:873] (1/4) Epoch 7, batch 2300, loss[loss=0.1336, simple_loss=0.1633, pruned_loss=0.05197, over 14697.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.1786, pruned_loss=0.07012, over 1978686.03 frames. ], batch size: 33, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:14:38,895 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5273, 1.6054, 2.8650, 2.0826, 2.6008, 1.7319, 2.1794, 2.5243], device='cuda:1'), covar=tensor([0.1045, 0.4435, 0.0374, 0.5215, 0.0557, 0.3183, 0.1412, 0.0589], device='cuda:1'), in_proj_covar=tensor([0.0222, 0.0237, 0.0165, 0.0320, 0.0189, 0.0245, 0.0224, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:15:05,643 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2022-12-07 16:15:07,552 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.381e+02 2.948e+02 3.680e+02 8.732e+02, threshold=5.896e+02, percent-clipped=3.0 2022-12-07 16:15:10,962 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:15:13,676 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:15:45,314 INFO [train.py:873] (1/4) Epoch 7, batch 2400, loss[loss=0.1475, simple_loss=0.1729, pruned_loss=0.06105, over 14393.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.1793, pruned_loss=0.07072, over 1981934.00 frames. ], batch size: 53, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:15:58,193 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:16:08,437 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7993, 5.2188, 5.2413, 5.7560, 5.4367, 4.5353, 5.7258, 4.7798], device='cuda:1'), covar=tensor([0.0269, 0.1049, 0.0297, 0.0357, 0.0570, 0.0338, 0.0464, 0.0428], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0234, 0.0150, 0.0145, 0.0153, 0.0122, 0.0228, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 16:16:08,563 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:16:17,327 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:16:36,796 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.462e+02 3.088e+02 3.941e+02 7.678e+02, threshold=6.176e+02, percent-clipped=7.0 2022-12-07 16:16:48,820 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0827, 1.3694, 3.7548, 1.6335, 3.9666, 4.0403, 3.0953, 4.4013], device='cuda:1'), covar=tensor([0.0195, 0.3226, 0.0452, 0.2415, 0.0354, 0.0384, 0.0548, 0.0128], device='cuda:1'), in_proj_covar=tensor([0.0162, 0.0161, 0.0150, 0.0167, 0.0161, 0.0159, 0.0130, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:16:53,146 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:17:11,972 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:17:14,361 INFO [train.py:873] (1/4) Epoch 7, batch 2500, loss[loss=0.1668, simple_loss=0.1565, pruned_loss=0.08853, over 1206.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.18, pruned_loss=0.07115, over 1995017.91 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:17:35,562 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:18:06,393 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.559e+02 3.205e+02 4.003e+02 7.455e+02, threshold=6.409e+02, percent-clipped=4.0 2022-12-07 16:18:18,479 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:18:27,613 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.45 vs. limit=5.0 2022-12-07 16:18:43,889 INFO [train.py:873] (1/4) Epoch 7, batch 2600, loss[loss=0.1868, simple_loss=0.1902, pruned_loss=0.09169, over 6030.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.1793, pruned_loss=0.07112, over 1885938.59 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:19:12,572 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-07 16:19:13,602 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 16:19:36,189 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.385e+02 3.008e+02 3.884e+02 8.359e+02, threshold=6.016e+02, percent-clipped=2.0 2022-12-07 16:19:40,163 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:13,150 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:13,912 INFO [train.py:873] (1/4) Epoch 7, batch 2700, loss[loss=0.2434, simple_loss=0.1996, pruned_loss=0.1436, over 1214.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.1787, pruned_loss=0.07076, over 1851909.43 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:20:23,345 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:23,475 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6775, 1.5967, 1.8622, 2.1192, 1.4399, 1.7195, 1.9768, 1.9263], device='cuda:1'), covar=tensor([0.0057, 0.0085, 0.0042, 0.0028, 0.0078, 0.0124, 0.0051, 0.0038], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0216, 0.0327, 0.0265, 0.0211, 0.0267, 0.0222, 0.0254], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:20:32,451 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2509, 1.0944, 1.1655, 1.0602, 1.3595, 0.6959, 1.2155, 1.4382], device='cuda:1'), covar=tensor([0.1110, 0.0905, 0.0417, 0.1410, 0.0821, 0.0731, 0.0793, 0.0555], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0019, 0.0019, 0.0019, 0.0019, 0.0027, 0.0020, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.9501e-05, 8.8646e-05, 8.8613e-05, 9.0036e-05, 8.8993e-05, 1.1607e-04, 9.4476e-05, 8.9118e-05], device='cuda:1') 2022-12-07 16:20:33,203 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:46,339 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0373, 3.7451, 3.4961, 3.6192, 3.8639, 3.8815, 3.9824, 3.9813], device='cuda:1'), covar=tensor([0.0745, 0.0586, 0.2055, 0.2664, 0.0675, 0.0683, 0.0939, 0.0802], device='cuda:1'), in_proj_covar=tensor([0.0327, 0.0234, 0.0396, 0.0491, 0.0285, 0.0356, 0.0351, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:21:06,887 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.439e+01 2.366e+02 3.097e+02 3.751e+02 6.035e+02, threshold=6.194e+02, percent-clipped=1.0 2022-12-07 16:21:08,857 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:21:16,304 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2613, 1.5349, 1.2360, 1.4788, 1.3876, 0.9376, 0.8236, 0.9939], device='cuda:1'), covar=tensor([0.0961, 0.1668, 0.1301, 0.0566, 0.1259, 0.0513, 0.0395, 0.0899], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0011, 0.0016, 0.0013, 0.0017], device='cuda:1'), out_proj_covar=tensor([7.2408e-05, 7.8665e-05, 7.1660e-05, 7.1768e-05, 7.4469e-05, 1.0487e-04, 8.9493e-05, 1.0121e-04], device='cuda:1') 2022-12-07 16:21:17,926 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:21:37,434 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:21:44,253 INFO [train.py:873] (1/4) Epoch 7, batch 2800, loss[loss=0.1949, simple_loss=0.1857, pruned_loss=0.102, over 3868.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1786, pruned_loss=0.07022, over 1902880.30 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:21:55,458 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0516, 2.0174, 1.9277, 2.1391, 1.8155, 1.9224, 2.0701, 2.0701], device='cuda:1'), covar=tensor([0.1091, 0.1084, 0.1071, 0.0838, 0.1302, 0.0924, 0.1173, 0.1063], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0106, 0.0120, 0.0124, 0.0124, 0.0096, 0.0136, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 16:22:02,680 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2022-12-07 16:22:20,025 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:22:36,454 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.244e+02 3.117e+02 3.930e+02 6.165e+02, threshold=6.234e+02, percent-clipped=0.0 2022-12-07 16:22:58,144 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 16:23:14,253 INFO [train.py:873] (1/4) Epoch 7, batch 2900, loss[loss=0.1511, simple_loss=0.1608, pruned_loss=0.07076, over 3869.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.1792, pruned_loss=0.07117, over 1920328.82 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:23:14,691 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:23:55,952 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3099, 2.1308, 2.6362, 1.6947, 1.7274, 2.3636, 1.2947, 2.1902], device='cuda:1'), covar=tensor([0.0897, 0.1772, 0.0848, 0.2452, 0.3120, 0.1007, 0.5283, 0.1352], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0086, 0.0079, 0.0087, 0.0110, 0.0070, 0.0128, 0.0076], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:24:01,606 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:24:07,740 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.377e+02 3.206e+02 3.983e+02 6.235e+02, threshold=6.412e+02, percent-clipped=1.0 2022-12-07 16:24:09,851 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2593, 1.2649, 1.1235, 1.4691, 1.3767, 1.1328, 1.0408, 0.8437], device='cuda:1'), covar=tensor([0.0747, 0.2333, 0.1355, 0.0762, 0.0835, 0.0456, 0.0407, 0.1414], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0012, 0.0017, 0.0013, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.4450e-05, 8.0526e-05, 7.3009e-05, 7.2528e-05, 7.6406e-05, 1.0809e-04, 9.3086e-05, 1.0392e-04], device='cuda:1') 2022-12-07 16:24:46,197 INFO [train.py:873] (1/4) Epoch 7, batch 3000, loss[loss=0.1445, simple_loss=0.1682, pruned_loss=0.06039, over 14406.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.1787, pruned_loss=0.0705, over 1962053.34 frames. ], batch size: 53, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:24:46,197 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 16:25:07,553 INFO [train.py:905] (1/4) Epoch 7, validation: loss=0.1228, simple_loss=0.1657, pruned_loss=0.03995, over 857387.00 frames. 2022-12-07 16:25:07,554 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 16:25:13,223 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9640, 2.4530, 3.9683, 4.2012, 4.1069, 2.5737, 4.1536, 3.3630], device='cuda:1'), covar=tensor([0.0191, 0.0506, 0.0516, 0.0238, 0.0156, 0.0737, 0.0169, 0.0447], device='cuda:1'), in_proj_covar=tensor([0.0229, 0.0217, 0.0330, 0.0266, 0.0213, 0.0267, 0.0223, 0.0251], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:25:19,616 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:27,572 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:30,245 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:54,439 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:58,959 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:01,391 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.229e+02 2.374e+02 2.963e+02 3.800e+02 1.337e+03, threshold=5.925e+02, percent-clipped=7.0 2022-12-07 16:26:10,818 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:12,855 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:18,883 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3348, 1.3857, 3.4371, 1.3855, 3.2679, 3.4702, 2.4317, 3.7435], device='cuda:1'), covar=tensor([0.0212, 0.2752, 0.0309, 0.2298, 0.0619, 0.0317, 0.0857, 0.0148], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0159, 0.0150, 0.0167, 0.0159, 0.0159, 0.0131, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:26:25,986 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:32,542 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:26:36,399 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 16:26:39,378 INFO [train.py:873] (1/4) Epoch 7, batch 3100, loss[loss=0.1771, simple_loss=0.1908, pruned_loss=0.08174, over 14389.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.1783, pruned_loss=0.07095, over 1905926.89 frames. ], batch size: 73, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:26:49,182 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:56,197 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:27:15,596 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:27:31,119 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.509e+02 3.187e+02 4.348e+02 7.597e+02, threshold=6.375e+02, percent-clipped=5.0 2022-12-07 16:27:46,277 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3958, 3.8011, 2.9199, 4.6129, 4.1566, 4.3505, 3.6332, 3.2159], device='cuda:1'), covar=tensor([0.0669, 0.1130, 0.3953, 0.0314, 0.0765, 0.0989, 0.1051, 0.3307], device='cuda:1'), in_proj_covar=tensor([0.0237, 0.0299, 0.0280, 0.0205, 0.0269, 0.0270, 0.0248, 0.0268], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:28:04,902 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:28:09,141 INFO [train.py:873] (1/4) Epoch 7, batch 3200, loss[loss=0.2329, simple_loss=0.1921, pruned_loss=0.1368, over 1202.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1782, pruned_loss=0.07037, over 1964332.25 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:28:43,665 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2022-12-07 16:29:02,498 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.245e+02 2.769e+02 3.901e+02 6.467e+02, threshold=5.537e+02, percent-clipped=1.0 2022-12-07 16:29:18,044 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.28 vs. limit=5.0 2022-12-07 16:29:40,602 INFO [train.py:873] (1/4) Epoch 7, batch 3300, loss[loss=0.1509, simple_loss=0.1712, pruned_loss=0.06536, over 14167.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.1775, pruned_loss=0.06961, over 1956255.96 frames. ], batch size: 84, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:29:46,913 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:30:04,212 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0351, 3.0156, 3.1846, 3.0728, 3.0754, 2.9374, 1.3720, 2.8983], device='cuda:1'), covar=tensor([0.0310, 0.0350, 0.0368, 0.0299, 0.0285, 0.0616, 0.2966, 0.0272], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0148, 0.0129, 0.0122, 0.0176, 0.0123, 0.0153, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:30:22,993 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2022-12-07 16:30:30,550 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:30:32,953 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.201e+02 2.834e+02 3.394e+02 6.106e+02, threshold=5.668e+02, percent-clipped=1.0 2022-12-07 16:30:53,232 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:02,745 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2196, 2.0724, 3.2473, 3.3584, 3.2465, 2.0707, 3.1873, 2.5351], device='cuda:1'), covar=tensor([0.0188, 0.0493, 0.0335, 0.0210, 0.0180, 0.0769, 0.0166, 0.0494], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0216, 0.0328, 0.0265, 0.0212, 0.0265, 0.0223, 0.0253], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:31:10,410 INFO [train.py:873] (1/4) Epoch 7, batch 3400, loss[loss=0.1534, simple_loss=0.1432, pruned_loss=0.08183, over 1244.00 frames. ], tot_loss[loss=0.1567, simple_loss=0.177, pruned_loss=0.06819, over 2013808.58 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:31:13,826 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:16,478 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:54,739 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:32:03,163 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2022-12-07 16:32:04,511 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.534e+01 2.454e+02 2.984e+02 3.725e+02 7.522e+02, threshold=5.968e+02, percent-clipped=2.0 2022-12-07 16:32:10,026 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:32:36,370 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:32:40,186 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 16:32:40,437 INFO [train.py:873] (1/4) Epoch 7, batch 3500, loss[loss=0.1511, simple_loss=0.1801, pruned_loss=0.06101, over 14316.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.1775, pruned_loss=0.06894, over 2000769.41 frames. ], batch size: 31, lr: 1.13e-02, grad_scale: 4.0 2022-12-07 16:32:48,510 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:33:03,310 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:33:18,182 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:33:31,827 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.610e+02 3.212e+02 4.326e+02 7.674e+02, threshold=6.425e+02, percent-clipped=4.0 2022-12-07 16:34:08,387 INFO [train.py:873] (1/4) Epoch 7, batch 3600, loss[loss=0.1651, simple_loss=0.1665, pruned_loss=0.08188, over 2662.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.1778, pruned_loss=0.06935, over 1962931.50 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:34:14,456 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:34:58,407 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:35:01,629 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.848e+01 2.321e+02 2.995e+02 3.518e+02 8.417e+02, threshold=5.990e+02, percent-clipped=3.0 2022-12-07 16:35:20,075 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:35:38,276 INFO [train.py:873] (1/4) Epoch 7, batch 3700, loss[loss=0.1879, simple_loss=0.1832, pruned_loss=0.09635, over 3903.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.1785, pruned_loss=0.06989, over 2003616.21 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:35:43,617 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:02,760 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:04,678 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2602, 1.3174, 1.4368, 1.2097, 1.4033, 0.6677, 1.4427, 1.5258], device='cuda:1'), covar=tensor([0.1490, 0.0672, 0.0501, 0.1397, 0.0935, 0.0514, 0.1287, 0.0687], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0020, 0.0019, 0.0027, 0.0020, 0.0020], device='cuda:1'), out_proj_covar=tensor([9.1735e-05, 9.1446e-05, 9.1495e-05, 9.4001e-05, 9.0154e-05, 1.1597e-04, 9.6450e-05, 9.3590e-05], device='cuda:1') 2022-12-07 16:36:26,228 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:30,493 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.323e+02 3.026e+02 3.847e+02 8.325e+02, threshold=6.052e+02, percent-clipped=4.0 2022-12-07 16:36:55,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9777, 3.7483, 3.2997, 2.6381, 3.1473, 3.5770, 3.9535, 3.1022], device='cuda:1'), covar=tensor([0.0544, 0.2382, 0.1297, 0.2160, 0.1157, 0.0639, 0.0819, 0.1569], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0196, 0.0126, 0.0130, 0.0117, 0.0118, 0.0097, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 16:37:07,567 INFO [train.py:873] (1/4) Epoch 7, batch 3800, loss[loss=0.1294, simple_loss=0.1644, pruned_loss=0.04725, over 14097.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.1777, pruned_loss=0.06947, over 1979721.50 frames. ], batch size: 29, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:37:11,433 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:37:12,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5363, 1.2699, 2.0701, 1.9298, 2.0215, 2.0259, 1.5539, 2.0527], device='cuda:1'), covar=tensor([0.0494, 0.0853, 0.0128, 0.0300, 0.0260, 0.0128, 0.0360, 0.0180], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0113, 0.0156, 0.0128, 0.0127, 0.0104, 0.0109], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:37:26,355 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:37:26,364 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1201, 2.0372, 1.7810, 1.7899, 2.0240, 2.0389, 2.0578, 2.0397], device='cuda:1'), covar=tensor([0.0998, 0.0880, 0.2586, 0.2883, 0.1210, 0.0947, 0.1409, 0.1220], device='cuda:1'), in_proj_covar=tensor([0.0331, 0.0237, 0.0393, 0.0499, 0.0290, 0.0360, 0.0352, 0.0316], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:38:01,122 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.147e+01 2.278e+02 2.831e+02 3.417e+02 7.148e+02, threshold=5.663e+02, percent-clipped=1.0 2022-12-07 16:38:04,779 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:38:37,956 INFO [train.py:873] (1/4) Epoch 7, batch 3900, loss[loss=0.1315, simple_loss=0.1624, pruned_loss=0.05029, over 13580.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.1773, pruned_loss=0.06882, over 1992075.45 frames. ], batch size: 17, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:38:41,377 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8285, 2.5749, 2.6535, 2.8489, 2.7659, 2.8098, 2.9259, 2.4098], device='cuda:1'), covar=tensor([0.0620, 0.1160, 0.0544, 0.0589, 0.0793, 0.0500, 0.0674, 0.0725], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0238, 0.0159, 0.0150, 0.0156, 0.0125, 0.0237, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 16:38:59,180 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:39:24,948 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7918, 1.5400, 4.2378, 4.1423, 3.9417, 4.4230, 4.0539, 4.3613], device='cuda:1'), covar=tensor([0.1897, 0.2074, 0.0205, 0.0218, 0.0272, 0.0206, 0.0192, 0.0221], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0159, 0.0113, 0.0155, 0.0129, 0.0128, 0.0105, 0.0110], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:39:30,001 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.072e+02 2.772e+02 3.353e+02 6.720e+02, threshold=5.544e+02, percent-clipped=1.0 2022-12-07 16:39:35,470 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.37 vs. limit=5.0 2022-12-07 16:39:41,387 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4639, 2.2508, 3.4058, 3.5662, 3.4764, 2.2315, 3.5144, 2.7010], device='cuda:1'), covar=tensor([0.0184, 0.0457, 0.0418, 0.0225, 0.0174, 0.0720, 0.0168, 0.0536], device='cuda:1'), in_proj_covar=tensor([0.0228, 0.0217, 0.0329, 0.0265, 0.0212, 0.0263, 0.0225, 0.0251], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 16:39:47,889 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-07 16:40:05,916 INFO [train.py:873] (1/4) Epoch 7, batch 4000, loss[loss=0.1459, simple_loss=0.1416, pruned_loss=0.07509, over 1193.00 frames. ], tot_loss[loss=0.1581, simple_loss=0.1779, pruned_loss=0.06917, over 2001611.23 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:40:12,931 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-07 16:40:37,828 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 16:40:49,534 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:40:50,972 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 16:40:58,796 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.814e+01 2.339e+02 2.890e+02 3.562e+02 6.238e+02, threshold=5.779e+02, percent-clipped=1.0 2022-12-07 16:40:59,040 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:41:26,827 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.5742, 0.3754, 0.4280, 0.4975, 0.5303, 0.3928, 0.5757, 0.4535], device='cuda:1'), covar=tensor([0.0231, 0.0254, 0.0147, 0.0200, 0.0330, 0.0650, 0.0348, 0.0574], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0011, 0.0012, 0.0017, 0.0014, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.6603e-05, 8.2015e-05, 7.5832e-05, 7.6583e-05, 7.8564e-05, 1.1453e-04, 9.5267e-05, 1.0873e-04], device='cuda:1') 2022-12-07 16:41:34,937 INFO [train.py:873] (1/4) Epoch 7, batch 4100, loss[loss=0.151, simple_loss=0.1751, pruned_loss=0.06344, over 14377.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.1783, pruned_loss=0.06958, over 2027363.99 frames. ], batch size: 31, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:41:38,581 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:41:43,962 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:41:52,456 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:41:53,177 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:42:20,799 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:42:26,657 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.128e+02 2.350e+02 2.777e+02 3.332e+02 5.900e+02, threshold=5.554e+02, percent-clipped=2.0 2022-12-07 16:42:35,648 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:43:02,444 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1052, 1.6893, 4.0876, 4.0012, 3.8979, 4.2579, 3.8032, 4.2720], device='cuda:1'), covar=tensor([0.1260, 0.1412, 0.0094, 0.0149, 0.0147, 0.0086, 0.0153, 0.0095], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0113, 0.0155, 0.0128, 0.0126, 0.0104, 0.0110], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:43:03,202 INFO [train.py:873] (1/4) Epoch 7, batch 4200, loss[loss=0.1336, simple_loss=0.1655, pruned_loss=0.05089, over 14266.00 frames. ], tot_loss[loss=0.159, simple_loss=0.1787, pruned_loss=0.06963, over 1972869.63 frames. ], batch size: 60, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:43:08,086 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8773, 1.3224, 3.1493, 2.9145, 3.0810, 3.1613, 2.6399, 3.1820], device='cuda:1'), covar=tensor([0.1148, 0.1355, 0.0111, 0.0246, 0.0179, 0.0125, 0.0261, 0.0138], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0113, 0.0156, 0.0129, 0.0126, 0.0104, 0.0111], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:43:20,487 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:43:57,174 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.513e+02 3.123e+02 4.078e+02 1.360e+03, threshold=6.245e+02, percent-clipped=14.0 2022-12-07 16:44:03,629 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 16:44:10,357 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2022-12-07 16:44:13,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5574, 3.0499, 4.1653, 3.0151, 4.2215, 4.2256, 3.9029, 3.5681], device='cuda:1'), covar=tensor([0.0522, 0.2664, 0.0869, 0.2201, 0.0757, 0.0768, 0.2567, 0.2231], device='cuda:1'), in_proj_covar=tensor([0.0309, 0.0321, 0.0378, 0.0306, 0.0361, 0.0304, 0.0357, 0.0333], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 16:44:32,551 INFO [train.py:873] (1/4) Epoch 7, batch 4300, loss[loss=0.175, simple_loss=0.1639, pruned_loss=0.09304, over 2635.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.1783, pruned_loss=0.06988, over 1915741.73 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:44:54,937 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9780, 2.0098, 1.9090, 2.0970, 1.7427, 1.9333, 2.0209, 2.0856], device='cuda:1'), covar=tensor([0.0977, 0.0987, 0.1096, 0.0932, 0.1370, 0.0845, 0.1079, 0.0845], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0104, 0.0119, 0.0123, 0.0122, 0.0094, 0.0133, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 16:45:14,773 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1858, 4.1225, 4.4500, 3.5227, 4.2486, 4.4496, 1.5431, 4.0246], device='cuda:1'), covar=tensor([0.0198, 0.0267, 0.0334, 0.0529, 0.0246, 0.0240, 0.3250, 0.0253], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0147, 0.0130, 0.0123, 0.0178, 0.0124, 0.0151, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:45:25,736 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.563e+02 3.565e+02 4.633e+02 7.506e+02, threshold=7.130e+02, percent-clipped=5.0 2022-12-07 16:45:35,281 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4977, 2.1862, 2.3266, 1.5153, 2.1723, 2.2530, 2.5745, 2.0967], device='cuda:1'), covar=tensor([0.0710, 0.1120, 0.0952, 0.2113, 0.0920, 0.0581, 0.0428, 0.1461], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0189, 0.0121, 0.0125, 0.0113, 0.0113, 0.0093, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 16:45:53,174 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.69 vs. limit=5.0 2022-12-07 16:45:56,353 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:02,014 INFO [train.py:873] (1/4) Epoch 7, batch 4400, loss[loss=0.1316, simple_loss=0.1703, pruned_loss=0.04644, over 14213.00 frames. ], tot_loss[loss=0.1589, simple_loss=0.1782, pruned_loss=0.06981, over 1946476.92 frames. ], batch size: 32, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:46:05,173 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7885, 1.5645, 1.8137, 1.9705, 1.5082, 1.7245, 1.9041, 1.7889], device='cuda:1'), covar=tensor([0.0070, 0.0151, 0.0061, 0.0044, 0.0134, 0.0185, 0.0097, 0.0059], device='cuda:1'), in_proj_covar=tensor([0.0228, 0.0220, 0.0331, 0.0268, 0.0214, 0.0265, 0.0228, 0.0252], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:46:06,657 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:15,370 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:46:25,260 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.35 vs. limit=2.0 2022-12-07 16:46:35,869 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2722, 3.7520, 2.9181, 4.5515, 4.0949, 4.2561, 3.5588, 3.1273], device='cuda:1'), covar=tensor([0.0747, 0.1334, 0.4048, 0.0322, 0.0935, 0.1539, 0.1221, 0.3726], device='cuda:1'), in_proj_covar=tensor([0.0241, 0.0300, 0.0283, 0.0208, 0.0270, 0.0271, 0.0249, 0.0268], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:46:40,097 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7860, 0.7081, 0.8507, 0.8049, 1.0004, 0.4098, 0.7287, 0.8060], device='cuda:1'), covar=tensor([0.0697, 0.0704, 0.0526, 0.0687, 0.0439, 0.0749, 0.0745, 0.0528], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0020, 0.0020, 0.0028, 0.0021, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.2011e-05, 9.2805e-05, 9.2760e-05, 9.3412e-05, 9.3500e-05, 1.1943e-04, 9.9105e-05, 9.6410e-05], device='cuda:1') 2022-12-07 16:46:50,844 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:55,943 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.417e+01 2.268e+02 2.875e+02 3.664e+02 5.837e+02, threshold=5.750e+02, percent-clipped=0.0 2022-12-07 16:47:16,458 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.21 vs. limit=5.0 2022-12-07 16:47:24,364 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8969, 1.4697, 3.2926, 2.9806, 3.1405, 3.2999, 2.6030, 3.3190], device='cuda:1'), covar=tensor([0.1047, 0.1252, 0.0093, 0.0250, 0.0192, 0.0114, 0.0321, 0.0118], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0156, 0.0111, 0.0155, 0.0128, 0.0125, 0.0104, 0.0109], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:47:31,485 INFO [train.py:873] (1/4) Epoch 7, batch 4500, loss[loss=0.1852, simple_loss=0.1869, pruned_loss=0.09169, over 7771.00 frames. ], tot_loss[loss=0.1585, simple_loss=0.178, pruned_loss=0.06955, over 1954201.04 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:47:48,519 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:47:54,019 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6220, 1.2937, 3.6769, 1.5315, 3.5336, 3.7099, 2.6474, 3.9655], device='cuda:1'), covar=tensor([0.0245, 0.3404, 0.0368, 0.2486, 0.0559, 0.0345, 0.0666, 0.0159], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0160, 0.0148, 0.0170, 0.0161, 0.0161, 0.0130, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:48:25,122 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.380e+02 2.362e+02 2.957e+02 3.544e+02 6.398e+02, threshold=5.915e+02, percent-clipped=1.0 2022-12-07 16:48:26,245 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8482, 1.7527, 3.1015, 2.3270, 2.9706, 1.7309, 2.5759, 2.8633], device='cuda:1'), covar=tensor([0.0704, 0.4561, 0.0405, 0.6456, 0.0598, 0.3884, 0.1228, 0.0518], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0239, 0.0174, 0.0323, 0.0194, 0.0243, 0.0228, 0.0185], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:48:31,735 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:49:00,839 INFO [train.py:873] (1/4) Epoch 7, batch 4600, loss[loss=0.1782, simple_loss=0.1859, pruned_loss=0.08526, over 7803.00 frames. ], tot_loss[loss=0.1581, simple_loss=0.1776, pruned_loss=0.06928, over 1958164.89 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:49:58,507 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.255e+02 2.301e+02 2.897e+02 3.880e+02 6.571e+02, threshold=5.793e+02, percent-clipped=2.0 2022-12-07 16:50:32,304 INFO [train.py:873] (1/4) Epoch 7, batch 4700, loss[loss=0.1708, simple_loss=0.181, pruned_loss=0.08033, over 5956.00 frames. ], tot_loss[loss=0.1571, simple_loss=0.1778, pruned_loss=0.06827, over 1997480.37 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:50:37,308 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:50:44,565 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.36 vs. limit=2.0 2022-12-07 16:50:45,879 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:50:53,443 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4102, 3.1045, 2.3554, 3.5235, 3.2813, 3.3345, 2.9175, 2.3389], device='cuda:1'), covar=tensor([0.0784, 0.1517, 0.4236, 0.0451, 0.0932, 0.1154, 0.1252, 0.3777], device='cuda:1'), in_proj_covar=tensor([0.0248, 0.0308, 0.0289, 0.0212, 0.0279, 0.0279, 0.0254, 0.0275], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:51:15,725 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:19,179 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:20,978 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:25,948 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.013e+02 2.271e+02 2.971e+02 3.641e+02 8.155e+02, threshold=5.943e+02, percent-clipped=3.0 2022-12-07 16:51:27,675 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:52:00,691 INFO [train.py:873] (1/4) Epoch 7, batch 4800, loss[loss=0.1406, simple_loss=0.1703, pruned_loss=0.05543, over 14175.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.1779, pruned_loss=0.06854, over 2001945.28 frames. ], batch size: 84, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:52:14,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0106, 2.0564, 1.8788, 2.1204, 1.7236, 1.8598, 2.0676, 2.0926], device='cuda:1'), covar=tensor([0.0860, 0.0889, 0.1044, 0.0751, 0.1220, 0.0813, 0.1043, 0.0743], device='cuda:1'), in_proj_covar=tensor([0.0115, 0.0107, 0.0121, 0.0125, 0.0125, 0.0096, 0.0134, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 16:52:14,765 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0233, 1.8574, 2.0690, 2.0067, 2.0761, 1.8481, 1.1953, 1.8406], device='cuda:1'), covar=tensor([0.0397, 0.0441, 0.0419, 0.0316, 0.0412, 0.0885, 0.1906, 0.0389], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0147, 0.0128, 0.0124, 0.0177, 0.0122, 0.0151, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 16:52:14,824 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:52:22,017 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:52:49,143 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=10.47 vs. limit=5.0 2022-12-07 16:52:50,175 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2022-12-07 16:52:54,460 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.403e+02 3.099e+02 4.047e+02 7.554e+02, threshold=6.197e+02, percent-clipped=4.0 2022-12-07 16:53:12,243 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2405, 2.1514, 4.2519, 2.9440, 4.0379, 2.0708, 3.0278, 3.8931], device='cuda:1'), covar=tensor([0.0409, 0.4977, 0.0297, 0.7475, 0.0435, 0.3578, 0.1415, 0.0310], device='cuda:1'), in_proj_covar=tensor([0.0222, 0.0239, 0.0175, 0.0323, 0.0195, 0.0245, 0.0228, 0.0184], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:53:15,934 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:53:29,364 INFO [train.py:873] (1/4) Epoch 7, batch 4900, loss[loss=0.1769, simple_loss=0.1963, pruned_loss=0.07877, over 14292.00 frames. ], tot_loss[loss=0.159, simple_loss=0.1786, pruned_loss=0.06968, over 1959014.02 frames. ], batch size: 35, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:54:23,753 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.354e+02 3.057e+02 3.894e+02 8.460e+02, threshold=6.115e+02, percent-clipped=5.0 2022-12-07 16:54:58,365 INFO [train.py:873] (1/4) Epoch 7, batch 5000, loss[loss=0.1583, simple_loss=0.1735, pruned_loss=0.07154, over 14156.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.1777, pruned_loss=0.06906, over 1985886.81 frames. ], batch size: 99, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:55:10,090 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3733, 1.8904, 3.5529, 2.5717, 3.3330, 1.8537, 2.6313, 3.3147], device='cuda:1'), covar=tensor([0.0725, 0.5110, 0.0461, 0.6684, 0.0553, 0.3948, 0.1465, 0.0401], device='cuda:1'), in_proj_covar=tensor([0.0225, 0.0239, 0.0177, 0.0325, 0.0195, 0.0247, 0.0230, 0.0184], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 16:55:42,207 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:55:52,924 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.513e+02 3.076e+02 4.134e+02 7.861e+02, threshold=6.153e+02, percent-clipped=5.0 2022-12-07 16:56:24,876 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:56:27,487 INFO [train.py:873] (1/4) Epoch 7, batch 5100, loss[loss=0.1349, simple_loss=0.1651, pruned_loss=0.05237, over 13964.00 frames. ], tot_loss[loss=0.1566, simple_loss=0.1771, pruned_loss=0.06809, over 2009096.63 frames. ], batch size: 26, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 16:56:37,629 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:56:46,205 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 16:57:20,715 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.452e+02 3.012e+02 3.839e+02 6.174e+02, threshold=6.024e+02, percent-clipped=1.0 2022-12-07 16:57:32,028 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:57:38,401 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 16:57:55,458 INFO [train.py:873] (1/4) Epoch 7, batch 5200, loss[loss=0.1496, simple_loss=0.178, pruned_loss=0.06058, over 14523.00 frames. ], tot_loss[loss=0.158, simple_loss=0.1775, pruned_loss=0.06923, over 1906752.66 frames. ], batch size: 34, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 16:57:55,671 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2131, 1.1946, 1.3559, 1.2447, 1.5897, 0.8770, 1.0440, 1.0104], device='cuda:1'), covar=tensor([0.0746, 0.2293, 0.0450, 0.0530, 0.0559, 0.0373, 0.0326, 0.0779], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0011, 0.0012, 0.0017, 0.0013, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.5045e-05, 8.1628e-05, 7.3227e-05, 7.6460e-05, 7.8375e-05, 1.1182e-04, 9.2622e-05, 1.0765e-04], device='cuda:1') 2022-12-07 16:58:23,654 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7921, 3.2446, 3.5852, 3.7904, 3.6269, 3.4831, 3.7845, 3.2887], device='cuda:1'), covar=tensor([0.0862, 0.1966, 0.0851, 0.0941, 0.1063, 0.1418, 0.1067, 0.0949], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0239, 0.0164, 0.0157, 0.0158, 0.0129, 0.0236, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 16:58:26,240 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:58:49,801 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.234e+02 2.881e+02 3.626e+02 9.172e+02, threshold=5.763e+02, percent-clipped=1.0 2022-12-07 16:59:24,243 INFO [train.py:873] (1/4) Epoch 7, batch 5300, loss[loss=0.2093, simple_loss=0.1799, pruned_loss=0.1193, over 1207.00 frames. ], tot_loss[loss=0.1563, simple_loss=0.1764, pruned_loss=0.06807, over 1943650.50 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:00:16,768 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 2.441e+02 3.172e+02 3.955e+02 6.877e+02, threshold=6.344e+02, percent-clipped=8.0 2022-12-07 17:00:49,949 INFO [train.py:873] (1/4) Epoch 7, batch 5400, loss[loss=0.159, simple_loss=0.1652, pruned_loss=0.07639, over 5975.00 frames. ], tot_loss[loss=0.1572, simple_loss=0.1773, pruned_loss=0.0686, over 1971352.32 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:00:55,234 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 17:00:58,017 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:00:59,909 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:01:05,876 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6582, 3.2696, 2.8908, 2.1004, 2.9781, 3.3040, 3.5293, 2.7380], device='cuda:1'), covar=tensor([0.0649, 0.2365, 0.1450, 0.2468, 0.1070, 0.0572, 0.1275, 0.1776], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0200, 0.0127, 0.0130, 0.0119, 0.0118, 0.0099, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 17:01:20,880 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:01:41,868 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:01:42,066 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1580, 2.7168, 3.7678, 3.0323, 3.9338, 3.7154, 3.7452, 3.2001], device='cuda:1'), covar=tensor([0.0427, 0.2485, 0.0946, 0.1658, 0.0675, 0.0739, 0.1402, 0.1660], device='cuda:1'), in_proj_covar=tensor([0.0308, 0.0325, 0.0379, 0.0304, 0.0363, 0.0303, 0.0355, 0.0328], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:01:43,629 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.469e+02 3.143e+02 3.850e+02 7.689e+02, threshold=6.287e+02, percent-clipped=2.0 2022-12-07 17:01:51,308 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:02:00,043 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:02:01,155 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-07 17:02:10,728 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9563, 3.6868, 3.7035, 4.0630, 3.6629, 3.2420, 4.0107, 4.0053], device='cuda:1'), covar=tensor([0.0754, 0.0705, 0.0735, 0.0555, 0.0815, 0.0685, 0.0729, 0.0664], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0108, 0.0124, 0.0126, 0.0125, 0.0096, 0.0138, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 17:02:14,151 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:02:17,732 INFO [train.py:873] (1/4) Epoch 7, batch 5500, loss[loss=0.2326, simple_loss=0.2029, pruned_loss=0.1311, over 1211.00 frames. ], tot_loss[loss=0.1555, simple_loss=0.176, pruned_loss=0.0675, over 1944659.68 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:02:42,259 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:02:43,071 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:02:50,587 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 17:03:10,733 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.288e+02 2.905e+02 3.483e+02 6.290e+02, threshold=5.809e+02, percent-clipped=1.0 2022-12-07 17:03:13,881 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2022-12-07 17:03:21,414 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:03:44,816 INFO [train.py:873] (1/4) Epoch 7, batch 5600, loss[loss=0.1903, simple_loss=0.1983, pruned_loss=0.09113, over 10338.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.1783, pruned_loss=0.06959, over 1982485.93 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:04:03,301 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1393, 3.3386, 2.7790, 4.3539, 4.0186, 4.1494, 3.4788, 2.7917], device='cuda:1'), covar=tensor([0.0633, 0.1545, 0.3921, 0.0420, 0.0776, 0.1059, 0.1085, 0.4192], device='cuda:1'), in_proj_covar=tensor([0.0241, 0.0300, 0.0281, 0.0207, 0.0275, 0.0272, 0.0253, 0.0271], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:04:15,891 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:04:38,633 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.390e+02 3.229e+02 4.054e+02 7.897e+02, threshold=6.457e+02, percent-clipped=5.0 2022-12-07 17:05:03,871 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:05:08,189 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1424, 2.0701, 3.1628, 3.2828, 3.1357, 2.1872, 3.1613, 2.3418], device='cuda:1'), covar=tensor([0.0154, 0.0460, 0.0389, 0.0207, 0.0174, 0.0666, 0.0157, 0.0512], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0221, 0.0328, 0.0273, 0.0216, 0.0263, 0.0232, 0.0255], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:05:13,365 INFO [train.py:873] (1/4) Epoch 7, batch 5700, loss[loss=0.2415, simple_loss=0.208, pruned_loss=0.1375, over 1211.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.1788, pruned_loss=0.07025, over 1954936.64 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:05:33,563 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2022-12-07 17:05:34,672 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2156, 1.9643, 2.0808, 2.1860, 2.1366, 2.0732, 2.2720, 1.7647], device='cuda:1'), covar=tensor([0.0644, 0.1407, 0.0663, 0.0789, 0.0849, 0.0632, 0.0838, 0.0704], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0231, 0.0160, 0.0151, 0.0156, 0.0123, 0.0230, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:05:56,823 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:06:05,990 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.301e+02 2.940e+02 3.681e+02 9.913e+02, threshold=5.881e+02, percent-clipped=3.0 2022-12-07 17:06:09,458 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:06:31,976 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:06:39,028 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:06:39,723 INFO [train.py:873] (1/4) Epoch 7, batch 5800, loss[loss=0.1877, simple_loss=0.1689, pruned_loss=0.1032, over 1210.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.1778, pruned_loss=0.0694, over 1939085.58 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:07:05,816 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:07:33,238 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 17:07:33,900 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.265e+02 2.888e+02 3.967e+02 8.048e+02, threshold=5.777e+02, percent-clipped=2.0 2022-12-07 17:07:39,256 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3827, 1.8492, 3.5589, 2.4504, 3.3268, 1.8142, 2.7347, 3.3254], device='cuda:1'), covar=tensor([0.0568, 0.4617, 0.0371, 0.5932, 0.0410, 0.3786, 0.1123, 0.0640], device='cuda:1'), in_proj_covar=tensor([0.0223, 0.0232, 0.0174, 0.0319, 0.0191, 0.0240, 0.0221, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:07:48,089 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:07:53,279 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4636, 1.1443, 1.3498, 0.9524, 1.0891, 1.4459, 1.2108, 1.1275], device='cuda:1'), covar=tensor([0.0219, 0.0452, 0.0295, 0.0313, 0.0756, 0.0380, 0.0278, 0.0762], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0194, 0.0123, 0.0126, 0.0118, 0.0117, 0.0098, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 17:08:00,200 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 17:08:08,702 INFO [train.py:873] (1/4) Epoch 7, batch 5900, loss[loss=0.1502, simple_loss=0.1749, pruned_loss=0.06272, over 14265.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.1767, pruned_loss=0.06859, over 1893221.57 frames. ], batch size: 46, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:08:09,313 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2022-12-07 17:08:11,925 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.27 vs. limit=5.0 2022-12-07 17:08:34,043 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:08:57,529 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5510, 4.3309, 4.1417, 4.6988, 4.2489, 3.9067, 4.6691, 4.6159], device='cuda:1'), covar=tensor([0.0622, 0.0661, 0.0696, 0.0559, 0.0694, 0.0596, 0.0521, 0.0616], device='cuda:1'), in_proj_covar=tensor([0.0119, 0.0112, 0.0124, 0.0129, 0.0127, 0.0099, 0.0140, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 17:09:01,666 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.784e+01 2.421e+02 2.907e+02 3.542e+02 5.776e+02, threshold=5.815e+02, percent-clipped=0.0 2022-12-07 17:09:06,084 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 17:09:35,644 INFO [train.py:873] (1/4) Epoch 7, batch 6000, loss[loss=0.1735, simple_loss=0.1834, pruned_loss=0.08184, over 6969.00 frames. ], tot_loss[loss=0.1566, simple_loss=0.1769, pruned_loss=0.06819, over 1974693.93 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:09:35,645 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 17:09:56,740 INFO [train.py:905] (1/4) Epoch 7, validation: loss=0.1227, simple_loss=0.1653, pruned_loss=0.04007, over 857387.00 frames. 2022-12-07 17:09:56,740 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 17:10:04,637 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:13,376 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:36,232 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:49,689 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 2.553e+02 3.284e+02 4.114e+02 1.252e+03, threshold=6.568e+02, percent-clipped=8.0 2022-12-07 17:10:53,243 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:57,489 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:05,516 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5099, 1.8918, 1.8226, 2.0135, 1.7810, 1.9921, 1.6419, 1.1329], device='cuda:1'), covar=tensor([0.1902, 0.1180, 0.0612, 0.0483, 0.1172, 0.0530, 0.1597, 0.3081], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0063, 0.0053, 0.0055, 0.0081, 0.0060, 0.0087, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 17:11:06,385 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:15,929 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:11:23,461 INFO [train.py:873] (1/4) Epoch 7, batch 6100, loss[loss=0.181, simple_loss=0.1593, pruned_loss=0.1014, over 1217.00 frames. ], tot_loss[loss=0.1572, simple_loss=0.1768, pruned_loss=0.06876, over 1899864.73 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:11:34,635 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:42,578 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:54,904 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.35 vs. limit=2.0 2022-12-07 17:11:57,326 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:12:10,865 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 17:12:16,319 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.483e+02 3.170e+02 3.695e+02 8.405e+02, threshold=6.340e+02, percent-clipped=2.0 2022-12-07 17:12:33,739 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.34 vs. limit=2.0 2022-12-07 17:12:35,860 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:12:50,517 INFO [train.py:873] (1/4) Epoch 7, batch 6200, loss[loss=0.172, simple_loss=0.1727, pruned_loss=0.08569, over 5959.00 frames. ], tot_loss[loss=0.1568, simple_loss=0.1766, pruned_loss=0.06852, over 1898491.95 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:13:02,944 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0728, 4.6421, 4.5876, 5.0989, 4.6450, 4.4467, 5.0847, 4.1716], device='cuda:1'), covar=tensor([0.0357, 0.0985, 0.0343, 0.0407, 0.0839, 0.0398, 0.0435, 0.0549], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0234, 0.0162, 0.0154, 0.0158, 0.0126, 0.0236, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:13:03,015 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9815, 1.8911, 4.2508, 3.9875, 4.0530, 4.3451, 3.8056, 4.4011], device='cuda:1'), covar=tensor([0.1219, 0.1234, 0.0086, 0.0147, 0.0137, 0.0095, 0.0159, 0.0096], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0115, 0.0157, 0.0129, 0.0128, 0.0105, 0.0111], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:13:07,076 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-07 17:13:16,598 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:13:44,670 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.329e+02 2.979e+02 3.715e+02 8.886e+02, threshold=5.958e+02, percent-clipped=2.0 2022-12-07 17:13:48,334 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6582, 1.8699, 2.6396, 2.6681, 2.6360, 1.9338, 2.7217, 2.2958], device='cuda:1'), covar=tensor([0.0181, 0.0450, 0.0271, 0.0237, 0.0215, 0.0651, 0.0152, 0.0448], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0220, 0.0328, 0.0270, 0.0215, 0.0263, 0.0229, 0.0254], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-07 17:13:58,643 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:14:19,069 INFO [train.py:873] (1/4) Epoch 7, batch 6300, loss[loss=0.1836, simple_loss=0.1811, pruned_loss=0.09306, over 4929.00 frames. ], tot_loss[loss=0.1556, simple_loss=0.1763, pruned_loss=0.06745, over 1956776.95 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:14:46,469 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3458, 3.1061, 2.3427, 3.5160, 3.3219, 3.3408, 2.8580, 2.4405], device='cuda:1'), covar=tensor([0.0866, 0.1459, 0.3562, 0.0449, 0.0884, 0.1009, 0.1377, 0.3775], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0300, 0.0280, 0.0207, 0.0270, 0.0269, 0.0252, 0.0264], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:14:49,146 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4358, 3.0522, 2.9552, 2.1731, 2.8383, 3.0479, 3.3992, 2.7627], device='cuda:1'), covar=tensor([0.0790, 0.2029, 0.1206, 0.2200, 0.1097, 0.0797, 0.1030, 0.1523], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0193, 0.0122, 0.0127, 0.0115, 0.0119, 0.0097, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 17:14:58,646 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:12,417 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.206e+02 2.744e+02 3.241e+02 7.232e+02, threshold=5.488e+02, percent-clipped=1.0 2022-12-07 17:15:16,224 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:23,962 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:24,660 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:40,152 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:46,557 INFO [train.py:873] (1/4) Epoch 7, batch 6400, loss[loss=0.1464, simple_loss=0.1751, pruned_loss=0.05886, over 14254.00 frames. ], tot_loss[loss=0.1541, simple_loss=0.1755, pruned_loss=0.06635, over 2010829.24 frames. ], batch size: 80, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:15:50,327 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9844, 1.4141, 2.7884, 2.6135, 2.7639, 2.7922, 2.1699, 2.8162], device='cuda:1'), covar=tensor([0.0741, 0.1085, 0.0107, 0.0258, 0.0232, 0.0105, 0.0325, 0.0146], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0114, 0.0156, 0.0130, 0.0128, 0.0105, 0.0112], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:16:17,637 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:16:35,066 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:16:39,967 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.371e+02 3.096e+02 4.074e+02 1.030e+03, threshold=6.192e+02, percent-clipped=3.0 2022-12-07 17:16:55,632 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:17:14,510 INFO [train.py:873] (1/4) Epoch 7, batch 6500, loss[loss=0.1502, simple_loss=0.1779, pruned_loss=0.06127, over 14609.00 frames. ], tot_loss[loss=0.1558, simple_loss=0.1766, pruned_loss=0.06749, over 2034493.72 frames. ], batch size: 22, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:17:17,361 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:18:07,971 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.306e+02 2.865e+02 3.658e+02 7.175e+02, threshold=5.730e+02, percent-clipped=1.0 2022-12-07 17:18:18,030 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1797, 2.2474, 2.9492, 2.5090, 2.9476, 2.9351, 2.9288, 2.5603], device='cuda:1'), covar=tensor([0.0695, 0.3025, 0.0998, 0.2264, 0.0677, 0.0823, 0.1623, 0.2205], device='cuda:1'), in_proj_covar=tensor([0.0307, 0.0328, 0.0381, 0.0305, 0.0366, 0.0297, 0.0351, 0.0328], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:18:38,873 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6636, 3.0017, 4.3686, 3.2931, 4.2847, 4.2614, 4.1107, 3.6768], device='cuda:1'), covar=tensor([0.0476, 0.2874, 0.1042, 0.1852, 0.0655, 0.0662, 0.1804, 0.2594], device='cuda:1'), in_proj_covar=tensor([0.0305, 0.0326, 0.0380, 0.0304, 0.0365, 0.0296, 0.0349, 0.0326], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:18:42,428 INFO [train.py:873] (1/4) Epoch 7, batch 6600, loss[loss=0.1443, simple_loss=0.1733, pruned_loss=0.05762, over 14186.00 frames. ], tot_loss[loss=0.1547, simple_loss=0.1758, pruned_loss=0.06675, over 2084341.58 frames. ], batch size: 84, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:19:00,056 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7607, 0.5363, 0.5981, 0.5043, 0.6490, 0.3045, 0.5666, 0.6468], device='cuda:1'), covar=tensor([0.0152, 0.0149, 0.0137, 0.0124, 0.0270, 0.0344, 0.0223, 0.0457], device='cuda:1'), in_proj_covar=tensor([0.0011, 0.0013, 0.0011, 0.0011, 0.0012, 0.0017, 0.0014, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.7940e-05, 8.4688e-05, 7.6080e-05, 7.8558e-05, 7.9564e-05, 1.1547e-04, 9.6418e-05, 1.1113e-04], device='cuda:1') 2022-12-07 17:19:14,563 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6055, 0.7275, 0.6414, 0.7400, 0.7678, 0.2076, 0.6830, 0.6612], device='cuda:1'), covar=tensor([0.0170, 0.0328, 0.0342, 0.0470, 0.0147, 0.0152, 0.0381, 0.0271], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0021, 0.0020, 0.0021, 0.0029, 0.0021, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.5374e-05, 9.7713e-05, 9.8807e-05, 9.6598e-05, 9.8030e-05, 1.2718e-04, 1.0129e-04, 9.9017e-05], device='cuda:1') 2022-12-07 17:19:35,920 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.181e+02 2.663e+02 3.409e+02 8.463e+02, threshold=5.326e+02, percent-clipped=3.0 2022-12-07 17:19:39,829 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:19:41,501 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:19:48,523 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:19:52,610 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 17:20:10,104 INFO [train.py:873] (1/4) Epoch 7, batch 6700, loss[loss=0.1441, simple_loss=0.1654, pruned_loss=0.06141, over 13881.00 frames. ], tot_loss[loss=0.1558, simple_loss=0.1763, pruned_loss=0.06766, over 2004689.95 frames. ], batch size: 23, lr: 1.10e-02, grad_scale: 16.0 2022-12-07 17:20:14,698 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3565, 2.0512, 2.5021, 1.5486, 1.7741, 2.2010, 1.1561, 2.0930], device='cuda:1'), covar=tensor([0.0735, 0.1640, 0.0799, 0.2272, 0.2942, 0.0970, 0.5140, 0.1046], device='cuda:1'), in_proj_covar=tensor([0.0074, 0.0086, 0.0079, 0.0087, 0.0108, 0.0072, 0.0127, 0.0079], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:20:21,600 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:30,295 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:34,937 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:36,543 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:21:03,676 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 2.435e+02 3.169e+02 4.506e+02 1.155e+03, threshold=6.339e+02, percent-clipped=9.0 2022-12-07 17:21:18,761 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:21:38,016 INFO [train.py:873] (1/4) Epoch 7, batch 6800, loss[loss=0.232, simple_loss=0.1867, pruned_loss=0.1387, over 1333.00 frames. ], tot_loss[loss=0.1557, simple_loss=0.1763, pruned_loss=0.06749, over 2007094.87 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 16.0 2022-12-07 17:21:55,396 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:22:00,778 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:22:06,172 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3230, 4.0350, 3.9481, 4.3419, 4.1142, 3.8253, 4.4001, 3.6971], device='cuda:1'), covar=tensor([0.0507, 0.0915, 0.0371, 0.0499, 0.0747, 0.1094, 0.0525, 0.0543], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0240, 0.0166, 0.0158, 0.0160, 0.0133, 0.0244, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:22:32,532 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.590e+02 3.090e+02 3.890e+02 7.235e+02, threshold=6.180e+02, percent-clipped=2.0 2022-12-07 17:22:49,504 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:23:05,760 INFO [train.py:873] (1/4) Epoch 7, batch 6900, loss[loss=0.1475, simple_loss=0.1764, pruned_loss=0.05924, over 14511.00 frames. ], tot_loss[loss=0.1553, simple_loss=0.176, pruned_loss=0.06729, over 1981052.70 frames. ], batch size: 22, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:23:16,209 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0259, 1.6721, 4.2022, 3.9642, 3.9616, 4.3326, 3.6825, 4.3223], device='cuda:1'), covar=tensor([0.1192, 0.1285, 0.0084, 0.0164, 0.0142, 0.0091, 0.0218, 0.0097], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0156, 0.0114, 0.0157, 0.0130, 0.0128, 0.0106, 0.0112], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:23:40,091 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:24:00,341 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.451e+02 3.077e+02 3.642e+02 5.911e+02, threshold=6.154e+02, percent-clipped=0.0 2022-12-07 17:24:04,526 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.46 vs. limit=2.0 2022-12-07 17:24:17,892 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.46 vs. limit=5.0 2022-12-07 17:24:33,250 INFO [train.py:873] (1/4) Epoch 7, batch 7000, loss[loss=0.1542, simple_loss=0.1834, pruned_loss=0.06248, over 14197.00 frames. ], tot_loss[loss=0.1557, simple_loss=0.1763, pruned_loss=0.06755, over 2022871.44 frames. ], batch size: 80, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:24:33,420 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:24:53,468 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:24:54,519 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9258, 2.0965, 4.0483, 2.6955, 3.7888, 2.1847, 3.0868, 3.7226], device='cuda:1'), covar=tensor([0.0489, 0.4826, 0.0325, 0.8264, 0.0556, 0.3706, 0.1278, 0.0444], device='cuda:1'), in_proj_covar=tensor([0.0229, 0.0237, 0.0179, 0.0322, 0.0199, 0.0247, 0.0229, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:25:00,110 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:25:27,917 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.412e+02 3.030e+02 3.769e+02 8.602e+02, threshold=6.059e+02, percent-clipped=2.0 2022-12-07 17:25:41,991 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:25:44,330 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 17:26:01,212 INFO [train.py:873] (1/4) Epoch 7, batch 7100, loss[loss=0.2105, simple_loss=0.1736, pruned_loss=0.1237, over 1267.00 frames. ], tot_loss[loss=0.1554, simple_loss=0.176, pruned_loss=0.06739, over 2011487.31 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:26:56,577 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.463e+02 3.010e+02 3.918e+02 9.338e+02, threshold=6.019e+02, percent-clipped=5.0 2022-12-07 17:26:57,989 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 17:27:08,061 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:27:13,425 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:27:29,108 INFO [train.py:873] (1/4) Epoch 7, batch 7200, loss[loss=0.1537, simple_loss=0.1757, pruned_loss=0.06584, over 14219.00 frames. ], tot_loss[loss=0.1573, simple_loss=0.1773, pruned_loss=0.0687, over 1958058.90 frames. ], batch size: 69, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:28:07,083 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:28:19,238 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5549, 1.1865, 2.8438, 2.7207, 2.8984, 2.8695, 1.9592, 2.7955], device='cuda:1'), covar=tensor([0.1640, 0.1946, 0.0235, 0.0393, 0.0318, 0.0243, 0.0630, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0157, 0.0116, 0.0159, 0.0130, 0.0128, 0.0106, 0.0114], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:28:24,357 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.010e+02 2.308e+02 2.919e+02 3.605e+02 5.197e+02, threshold=5.838e+02, percent-clipped=0.0 2022-12-07 17:28:47,668 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3225, 1.3525, 1.1428, 1.4044, 1.4300, 1.5557, 1.3196, 1.4107], device='cuda:1'), covar=tensor([0.0630, 0.1268, 0.1089, 0.0727, 0.1554, 0.0431, 0.0437, 0.0651], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0014, 0.0018], device='cuda:1'), out_proj_covar=tensor([7.9748e-05, 8.4823e-05, 7.6747e-05, 7.7978e-05, 8.1281e-05, 1.1781e-04, 9.7129e-05, 1.1187e-04], device='cuda:1') 2022-12-07 17:28:52,712 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:28:56,838 INFO [train.py:873] (1/4) Epoch 7, batch 7300, loss[loss=0.1967, simple_loss=0.1659, pruned_loss=0.1138, over 1287.00 frames. ], tot_loss[loss=0.1561, simple_loss=0.1761, pruned_loss=0.06807, over 1901066.76 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:29:16,727 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:29:22,723 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 17:29:38,887 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9554, 2.8092, 2.0425, 3.0765, 2.8412, 2.8925, 2.4589, 2.2425], device='cuda:1'), covar=tensor([0.0701, 0.1471, 0.4239, 0.0480, 0.0995, 0.0789, 0.1677, 0.3118], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0299, 0.0284, 0.0213, 0.0275, 0.0267, 0.0254, 0.0270], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:29:51,339 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.341e+02 3.110e+02 4.071e+02 1.233e+03, threshold=6.219e+02, percent-clipped=5.0 2022-12-07 17:29:58,693 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:30:05,537 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:30:24,391 INFO [train.py:873] (1/4) Epoch 7, batch 7400, loss[loss=0.1545, simple_loss=0.1731, pruned_loss=0.06798, over 14207.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.1759, pruned_loss=0.06839, over 1867195.69 frames. ], batch size: 37, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:30:24,875 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-07 17:30:29,249 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-07 17:30:45,435 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3982, 3.5415, 4.3220, 2.9858, 2.7930, 3.8037, 1.9811, 3.4996], device='cuda:1'), covar=tensor([0.1256, 0.0756, 0.0695, 0.1884, 0.2351, 0.0845, 0.5033, 0.1003], device='cuda:1'), in_proj_covar=tensor([0.0076, 0.0090, 0.0082, 0.0091, 0.0112, 0.0075, 0.0134, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:31:00,172 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:31:19,607 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.081e+01 2.208e+02 2.890e+02 3.954e+02 8.452e+02, threshold=5.779e+02, percent-clipped=3.0 2022-12-07 17:31:31,467 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:31:52,180 INFO [train.py:873] (1/4) Epoch 7, batch 7500, loss[loss=0.1637, simple_loss=0.1874, pruned_loss=0.07, over 14082.00 frames. ], tot_loss[loss=0.1557, simple_loss=0.1757, pruned_loss=0.06785, over 1920821.43 frames. ], batch size: 29, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:32:13,105 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:32:24,642 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:33:18,931 INFO [train.py:873] (1/4) Epoch 8, batch 0, loss[loss=0.1927, simple_loss=0.198, pruned_loss=0.09369, over 10344.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.198, pruned_loss=0.09369, over 10344.00 frames. ], batch size: 100, lr: 1.03e-02, grad_scale: 8.0 2022-12-07 17:33:18,931 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 17:33:26,263 INFO [train.py:905] (1/4) Epoch 8, validation: loss=0.1282, simple_loss=0.1716, pruned_loss=0.04242, over 857387.00 frames. 2022-12-07 17:33:26,263 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 17:33:27,146 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.973e+01 1.673e+02 2.697e+02 3.774e+02 7.847e+02, threshold=5.395e+02, percent-clipped=3.0 2022-12-07 17:33:44,408 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 17:33:56,357 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:34:38,737 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:34:55,991 INFO [train.py:873] (1/4) Epoch 8, batch 100, loss[loss=0.172, simple_loss=0.1854, pruned_loss=0.07931, over 13528.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.1781, pruned_loss=0.06866, over 863357.64 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:34:56,725 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.496e+02 3.096e+02 4.127e+02 8.461e+02, threshold=6.192e+02, percent-clipped=10.0 2022-12-07 17:35:06,694 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1253, 3.8065, 3.6606, 4.1222, 3.8687, 3.6330, 4.1661, 3.4576], device='cuda:1'), covar=tensor([0.0464, 0.0900, 0.0422, 0.0444, 0.0775, 0.1036, 0.0575, 0.0556], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0240, 0.0164, 0.0155, 0.0158, 0.0128, 0.0243, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:35:13,436 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:35:46,747 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8745, 3.8023, 3.8119, 3.3769, 3.8024, 3.9416, 1.3323, 3.5962], device='cuda:1'), covar=tensor([0.0331, 0.0488, 0.0737, 0.0739, 0.0494, 0.0496, 0.4241, 0.0395], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0147, 0.0129, 0.0124, 0.0179, 0.0121, 0.0151, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:35:59,492 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:36:03,903 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-07 17:36:06,744 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4849, 3.2656, 3.2098, 3.5660, 3.0797, 2.9835, 3.5093, 3.5330], device='cuda:1'), covar=tensor([0.0757, 0.0787, 0.0886, 0.0649, 0.0987, 0.0726, 0.0661, 0.0796], device='cuda:1'), in_proj_covar=tensor([0.0118, 0.0110, 0.0120, 0.0126, 0.0127, 0.0097, 0.0137, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 17:36:07,748 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:36:12,911 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7441, 3.3025, 2.5623, 3.8303, 3.6836, 3.7082, 3.1526, 2.4912], device='cuda:1'), covar=tensor([0.0616, 0.1385, 0.4040, 0.0424, 0.0660, 0.0901, 0.1288, 0.4131], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0297, 0.0281, 0.0212, 0.0275, 0.0269, 0.0251, 0.0270], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:36:23,090 INFO [train.py:873] (1/4) Epoch 8, batch 200, loss[loss=0.1581, simple_loss=0.173, pruned_loss=0.07164, over 6988.00 frames. ], tot_loss[loss=0.1552, simple_loss=0.1768, pruned_loss=0.06679, over 1383768.70 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:36:23,899 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.183e+02 2.918e+02 3.622e+02 1.019e+03, threshold=5.836e+02, percent-clipped=2.0 2022-12-07 17:36:32,577 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8530, 4.4867, 5.0180, 4.0903, 4.6704, 5.3400, 1.7739, 4.4796], device='cuda:1'), covar=tensor([0.0219, 0.0493, 0.0422, 0.0609, 0.0494, 0.0103, 0.3515, 0.0276], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0148, 0.0129, 0.0124, 0.0180, 0.0121, 0.0151, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:36:36,233 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6115, 1.1994, 2.0584, 1.8834, 1.9367, 2.0679, 1.5361, 2.0519], device='cuda:1'), covar=tensor([0.0594, 0.0943, 0.0122, 0.0312, 0.0328, 0.0132, 0.0419, 0.0190], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0157, 0.0116, 0.0160, 0.0132, 0.0129, 0.0108, 0.0114], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:36:39,837 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2022-12-07 17:36:52,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0506, 2.0897, 1.9705, 2.2083, 1.6810, 2.0321, 2.1177, 2.1536], device='cuda:1'), covar=tensor([0.0937, 0.0930, 0.0966, 0.0647, 0.1392, 0.0690, 0.0957, 0.0790], device='cuda:1'), in_proj_covar=tensor([0.0116, 0.0109, 0.0119, 0.0124, 0.0125, 0.0096, 0.0135, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 17:37:30,592 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:37:49,174 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-07 17:37:52,051 INFO [train.py:873] (1/4) Epoch 8, batch 300, loss[loss=0.138, simple_loss=0.1662, pruned_loss=0.05486, over 14274.00 frames. ], tot_loss[loss=0.1537, simple_loss=0.1752, pruned_loss=0.06614, over 1652951.32 frames. ], batch size: 25, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:37:52,977 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.915e+01 2.253e+02 2.827e+02 3.791e+02 8.401e+02, threshold=5.654e+02, percent-clipped=6.0 2022-12-07 17:37:57,644 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:38:12,814 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:38:20,883 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6745, 2.6585, 2.7622, 2.7289, 2.6534, 2.3786, 1.3844, 2.5008], device='cuda:1'), covar=tensor([0.0346, 0.0395, 0.0425, 0.0364, 0.0359, 0.1071, 0.2623, 0.0330], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0150, 0.0130, 0.0126, 0.0182, 0.0122, 0.0153, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:38:49,105 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4464, 5.2545, 5.0184, 5.5456, 5.0028, 4.6533, 5.5810, 5.3800], device='cuda:1'), covar=tensor([0.0708, 0.0419, 0.0583, 0.0446, 0.0653, 0.0404, 0.0397, 0.0647], device='cuda:1'), in_proj_covar=tensor([0.0119, 0.0111, 0.0122, 0.0126, 0.0126, 0.0097, 0.0137, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 17:38:50,116 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:38:51,778 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:38:54,758 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9322, 4.7044, 4.3232, 4.4460, 4.5325, 4.7538, 4.8967, 4.9182], device='cuda:1'), covar=tensor([0.0802, 0.0449, 0.2680, 0.3114, 0.0801, 0.0747, 0.1144, 0.0859], device='cuda:1'), in_proj_covar=tensor([0.0330, 0.0235, 0.0398, 0.0498, 0.0289, 0.0362, 0.0362, 0.0314], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:39:02,122 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:18,813 INFO [train.py:873] (1/4) Epoch 8, batch 400, loss[loss=0.2431, simple_loss=0.2034, pruned_loss=0.1414, over 1230.00 frames. ], tot_loss[loss=0.154, simple_loss=0.175, pruned_loss=0.06646, over 1771130.54 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:39:19,679 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.324e+02 2.854e+02 3.563e+02 7.977e+02, threshold=5.709e+02, percent-clipped=8.0 2022-12-07 17:39:24,304 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8226, 3.8375, 4.1298, 3.5611, 3.8815, 4.0075, 1.4066, 3.6633], device='cuda:1'), covar=tensor([0.0237, 0.0304, 0.0318, 0.0419, 0.0313, 0.0319, 0.3281, 0.0281], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0149, 0.0129, 0.0125, 0.0180, 0.0122, 0.0152, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:39:45,415 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:47,065 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:54,876 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:23,193 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:26,502 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:29,252 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9052, 1.4376, 3.0955, 2.8677, 3.0241, 3.1079, 2.5469, 3.1364], device='cuda:1'), covar=tensor([0.1054, 0.1212, 0.0106, 0.0264, 0.0207, 0.0115, 0.0303, 0.0129], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0155, 0.0116, 0.0159, 0.0132, 0.0129, 0.0107, 0.0114], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:40:40,771 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 17:40:45,937 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2022-12-07 17:40:47,040 INFO [train.py:873] (1/4) Epoch 8, batch 500, loss[loss=0.1441, simple_loss=0.1663, pruned_loss=0.06088, over 14283.00 frames. ], tot_loss[loss=0.1518, simple_loss=0.1736, pruned_loss=0.06501, over 1877547.83 frames. ], batch size: 63, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:40:48,831 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.493e+02 3.208e+02 4.073e+02 9.438e+02, threshold=6.416e+02, percent-clipped=6.0 2022-12-07 17:41:05,151 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:42:13,824 INFO [train.py:873] (1/4) Epoch 8, batch 600, loss[loss=0.138, simple_loss=0.1624, pruned_loss=0.05681, over 11179.00 frames. ], tot_loss[loss=0.1529, simple_loss=0.1741, pruned_loss=0.06583, over 1880697.35 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:42:15,675 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.499e+01 2.137e+02 2.538e+02 3.360e+02 7.193e+02, threshold=5.075e+02, percent-clipped=1.0 2022-12-07 17:42:45,105 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4264, 4.8202, 4.7611, 5.3034, 4.9995, 4.5454, 5.2779, 4.2391], device='cuda:1'), covar=tensor([0.0287, 0.1026, 0.0271, 0.0362, 0.0768, 0.0354, 0.0436, 0.0575], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0241, 0.0166, 0.0155, 0.0161, 0.0129, 0.0243, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:43:08,740 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:43:42,308 INFO [train.py:873] (1/4) Epoch 8, batch 700, loss[loss=0.1804, simple_loss=0.1947, pruned_loss=0.08304, over 9467.00 frames. ], tot_loss[loss=0.1525, simple_loss=0.1733, pruned_loss=0.06579, over 1858969.89 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:43:44,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.400e+02 3.018e+02 4.076e+02 1.039e+03, threshold=6.035e+02, percent-clipped=12.0 2022-12-07 17:43:53,134 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7386, 4.7270, 5.0489, 4.3776, 4.7184, 5.3170, 1.8617, 4.4032], device='cuda:1'), covar=tensor([0.0258, 0.0323, 0.0419, 0.0513, 0.0409, 0.0101, 0.3532, 0.0335], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0150, 0.0130, 0.0126, 0.0181, 0.0122, 0.0153, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:43:57,568 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1481, 3.8790, 3.5875, 3.7311, 3.9178, 4.0009, 4.1249, 4.0741], device='cuda:1'), covar=tensor([0.0787, 0.0570, 0.2502, 0.2808, 0.0892, 0.0869, 0.1008, 0.0821], device='cuda:1'), in_proj_covar=tensor([0.0333, 0.0237, 0.0404, 0.0501, 0.0295, 0.0369, 0.0365, 0.0317], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:44:02,734 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8713, 3.6563, 3.4003, 3.4842, 3.7260, 3.7382, 3.8867, 3.8249], device='cuda:1'), covar=tensor([0.0912, 0.0607, 0.2262, 0.2988, 0.0869, 0.0946, 0.1025, 0.0826], device='cuda:1'), in_proj_covar=tensor([0.0332, 0.0237, 0.0404, 0.0501, 0.0295, 0.0369, 0.0364, 0.0317], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:44:04,137 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:13,800 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:20,233 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5590, 2.3440, 3.0418, 1.8005, 1.9769, 2.5277, 1.3112, 2.4737], device='cuda:1'), covar=tensor([0.1148, 0.1186, 0.0542, 0.2237, 0.2570, 0.1308, 0.4904, 0.1121], device='cuda:1'), in_proj_covar=tensor([0.0074, 0.0085, 0.0078, 0.0085, 0.0107, 0.0072, 0.0127, 0.0077], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:44:49,968 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:59,974 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 17:45:10,196 INFO [train.py:873] (1/4) Epoch 8, batch 800, loss[loss=0.1612, simple_loss=0.1837, pruned_loss=0.0693, over 13954.00 frames. ], tot_loss[loss=0.1532, simple_loss=0.174, pruned_loss=0.06622, over 1926949.50 frames. ], batch size: 23, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:45:11,875 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.469e+01 2.335e+02 2.764e+02 3.237e+02 6.593e+02, threshold=5.529e+02, percent-clipped=1.0 2022-12-07 17:45:32,804 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:45:32,862 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9431, 5.2664, 5.2473, 5.8435, 5.4984, 4.7192, 5.7802, 4.6267], device='cuda:1'), covar=tensor([0.0269, 0.1118, 0.0294, 0.0384, 0.0624, 0.0367, 0.0519, 0.0540], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0239, 0.0166, 0.0155, 0.0160, 0.0129, 0.0244, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:45:33,732 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8774, 2.6024, 2.6783, 2.8286, 2.7772, 2.7909, 2.9613, 2.4054], device='cuda:1'), covar=tensor([0.0676, 0.1248, 0.0600, 0.0660, 0.0833, 0.0588, 0.0725, 0.0818], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0239, 0.0166, 0.0155, 0.0160, 0.0130, 0.0244, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 17:45:42,622 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=9.33 vs. limit=5.0 2022-12-07 17:46:23,130 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5955, 1.9183, 2.4881, 2.1724, 2.4999, 2.3234, 2.3285, 2.3558], device='cuda:1'), covar=tensor([0.0438, 0.2507, 0.0577, 0.1374, 0.0377, 0.0875, 0.0710, 0.1510], device='cuda:1'), in_proj_covar=tensor([0.0318, 0.0330, 0.0384, 0.0310, 0.0375, 0.0304, 0.0359, 0.0332], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:46:28,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1221, 1.4788, 1.7330, 1.4577, 1.2264, 1.2845, 1.1320, 1.1077], device='cuda:1'), covar=tensor([0.0595, 0.0955, 0.0698, 0.0446, 0.1080, 0.0510, 0.0634, 0.0924], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0015, 0.0018], device='cuda:1'), out_proj_covar=tensor([8.3144e-05, 8.7770e-05, 7.8228e-05, 8.3306e-05, 8.2310e-05, 1.2003e-04, 1.0280e-04, 1.1385e-04], device='cuda:1') 2022-12-07 17:46:38,790 INFO [train.py:873] (1/4) Epoch 8, batch 900, loss[loss=0.1513, simple_loss=0.1725, pruned_loss=0.065, over 14253.00 frames. ], tot_loss[loss=0.1538, simple_loss=0.1745, pruned_loss=0.0665, over 1960440.39 frames. ], batch size: 37, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:46:40,850 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.164e+02 2.734e+02 3.492e+02 7.190e+02, threshold=5.467e+02, percent-clipped=1.0 2022-12-07 17:46:49,388 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4169, 1.0340, 1.3769, 0.8622, 0.9742, 1.3993, 1.2306, 1.1926], device='cuda:1'), covar=tensor([0.0235, 0.0741, 0.0438, 0.0461, 0.0728, 0.0512, 0.0258, 0.0892], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0189, 0.0123, 0.0125, 0.0116, 0.0121, 0.0099, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 17:47:11,774 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8574, 1.3445, 2.0555, 1.2088, 1.9984, 2.0900, 1.7480, 2.0976], device='cuda:1'), covar=tensor([0.0253, 0.1696, 0.0360, 0.1790, 0.0422, 0.0368, 0.0876, 0.0307], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0161, 0.0153, 0.0169, 0.0164, 0.0164, 0.0132, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:47:32,880 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:04,457 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2022-12-07 17:48:05,623 INFO [train.py:873] (1/4) Epoch 8, batch 1000, loss[loss=0.1489, simple_loss=0.1639, pruned_loss=0.06695, over 6917.00 frames. ], tot_loss[loss=0.1527, simple_loss=0.1741, pruned_loss=0.06571, over 1979175.79 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:48:06,964 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 17:48:07,330 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.511e+01 2.177e+02 2.724e+02 3.766e+02 8.147e+02, threshold=5.449e+02, percent-clipped=5.0 2022-12-07 17:48:14,337 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:27,846 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:36,600 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3532, 2.0921, 4.3447, 2.9632, 4.1790, 1.9312, 3.1908, 4.0224], device='cuda:1'), covar=tensor([0.0400, 0.4897, 0.0370, 0.7810, 0.0444, 0.3976, 0.1282, 0.0502], device='cuda:1'), in_proj_covar=tensor([0.0227, 0.0234, 0.0181, 0.0321, 0.0202, 0.0244, 0.0229, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:48:37,358 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:59,668 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5821, 1.8156, 2.3492, 1.9880, 2.5120, 2.4048, 2.3614, 2.1881], device='cuda:1'), covar=tensor([0.0446, 0.2696, 0.0573, 0.1736, 0.0448, 0.0981, 0.0739, 0.2094], device='cuda:1'), in_proj_covar=tensor([0.0319, 0.0327, 0.0383, 0.0310, 0.0375, 0.0304, 0.0359, 0.0331], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:49:09,284 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:49:12,080 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0011, 3.4941, 2.6647, 4.1891, 3.9664, 3.8914, 3.5405, 2.7986], device='cuda:1'), covar=tensor([0.0786, 0.1235, 0.4108, 0.0391, 0.0759, 0.1442, 0.1010, 0.4054], device='cuda:1'), in_proj_covar=tensor([0.0249, 0.0298, 0.0281, 0.0213, 0.0279, 0.0276, 0.0251, 0.0272], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:49:18,693 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:49:22,077 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:49:32,902 INFO [train.py:873] (1/4) Epoch 8, batch 1100, loss[loss=0.1684, simple_loss=0.1559, pruned_loss=0.09046, over 2602.00 frames. ], tot_loss[loss=0.1524, simple_loss=0.1735, pruned_loss=0.06561, over 1908529.46 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:49:34,931 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.125e+02 2.772e+02 3.608e+02 1.248e+03, threshold=5.543e+02, percent-clipped=5.0 2022-12-07 17:49:53,979 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1294, 1.9081, 3.3243, 2.3688, 3.1322, 1.8539, 2.5732, 3.1305], device='cuda:1'), covar=tensor([0.0871, 0.4797, 0.0434, 0.6159, 0.0562, 0.3593, 0.1387, 0.0463], device='cuda:1'), in_proj_covar=tensor([0.0228, 0.0235, 0.0180, 0.0317, 0.0203, 0.0241, 0.0228, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:50:04,026 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:50:35,485 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8561, 1.2317, 2.9820, 2.7622, 2.9188, 3.0149, 2.3842, 3.0005], device='cuda:1'), covar=tensor([0.1065, 0.1397, 0.0162, 0.0321, 0.0282, 0.0158, 0.0360, 0.0186], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0156, 0.0115, 0.0159, 0.0132, 0.0129, 0.0107, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:50:49,997 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.55 vs. limit=5.0 2022-12-07 17:50:59,597 INFO [train.py:873] (1/4) Epoch 8, batch 1200, loss[loss=0.1499, simple_loss=0.175, pruned_loss=0.06241, over 14294.00 frames. ], tot_loss[loss=0.1518, simple_loss=0.1734, pruned_loss=0.0651, over 1986484.12 frames. ], batch size: 63, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:51:02,201 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.508e+02 3.088e+02 3.577e+02 7.086e+02, threshold=6.177e+02, percent-clipped=4.0 2022-12-07 17:51:42,142 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2022-12-07 17:51:43,393 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6520, 2.5426, 2.7527, 2.6974, 2.6208, 2.4377, 1.3745, 2.3992], device='cuda:1'), covar=tensor([0.0351, 0.0406, 0.0445, 0.0357, 0.0367, 0.0831, 0.2408, 0.0357], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0148, 0.0130, 0.0125, 0.0179, 0.0121, 0.0150, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:52:06,948 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-07 17:52:24,290 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2022-12-07 17:52:27,388 INFO [train.py:873] (1/4) Epoch 8, batch 1300, loss[loss=0.1568, simple_loss=0.1729, pruned_loss=0.07037, over 5960.00 frames. ], tot_loss[loss=0.1518, simple_loss=0.1733, pruned_loss=0.06514, over 1932488.93 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:52:30,351 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.196e+02 2.858e+02 3.785e+02 8.723e+02, threshold=5.716e+02, percent-clipped=2.0 2022-12-07 17:53:22,040 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2302, 1.3797, 2.5133, 1.2458, 2.4781, 2.4532, 1.9506, 2.4434], device='cuda:1'), covar=tensor([0.0297, 0.2080, 0.0265, 0.1880, 0.0351, 0.0436, 0.0910, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0159, 0.0151, 0.0168, 0.0164, 0.0164, 0.0133, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:53:39,924 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 17:53:54,867 INFO [train.py:873] (1/4) Epoch 8, batch 1400, loss[loss=0.1505, simple_loss=0.1739, pruned_loss=0.06355, over 14260.00 frames. ], tot_loss[loss=0.1516, simple_loss=0.1734, pruned_loss=0.06492, over 1956900.85 frames. ], batch size: 28, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:53:57,393 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.292e+02 2.799e+02 3.377e+02 6.597e+02, threshold=5.599e+02, percent-clipped=2.0 2022-12-07 17:54:25,970 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=10.13 vs. limit=5.0 2022-12-07 17:54:27,298 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4622, 2.3627, 2.5377, 2.5040, 2.4541, 2.1508, 1.4717, 2.2046], device='cuda:1'), covar=tensor([0.0367, 0.0439, 0.0409, 0.0360, 0.0325, 0.1141, 0.2129, 0.0371], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0150, 0.0133, 0.0127, 0.0181, 0.0125, 0.0153, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:54:31,752 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5946, 4.0318, 3.1849, 4.8363, 4.3723, 4.6660, 3.7784, 3.5381], device='cuda:1'), covar=tensor([0.0800, 0.1219, 0.4188, 0.0413, 0.0991, 0.1201, 0.1209, 0.3585], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0293, 0.0273, 0.0212, 0.0273, 0.0269, 0.0247, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:55:19,622 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0312, 1.8201, 4.3571, 4.0470, 4.0740, 4.5405, 4.0750, 4.4616], device='cuda:1'), covar=tensor([0.1242, 0.1312, 0.0083, 0.0149, 0.0132, 0.0071, 0.0137, 0.0107], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0154, 0.0113, 0.0158, 0.0131, 0.0127, 0.0106, 0.0111], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:55:22,064 INFO [train.py:873] (1/4) Epoch 8, batch 1500, loss[loss=0.1553, simple_loss=0.1707, pruned_loss=0.06991, over 8602.00 frames. ], tot_loss[loss=0.1513, simple_loss=0.1731, pruned_loss=0.06476, over 1927769.61 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:55:25,014 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.978e+01 2.195e+02 2.660e+02 3.596e+02 7.442e+02, threshold=5.321e+02, percent-clipped=3.0 2022-12-07 17:55:37,255 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6819, 2.6600, 1.9671, 2.7946, 2.5437, 2.6517, 2.2754, 2.1917], device='cuda:1'), covar=tensor([0.0678, 0.0995, 0.2967, 0.0520, 0.1035, 0.0822, 0.1476, 0.2631], device='cuda:1'), in_proj_covar=tensor([0.0244, 0.0294, 0.0274, 0.0212, 0.0273, 0.0271, 0.0248, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:56:02,247 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5768, 3.3704, 3.0049, 2.2626, 2.9595, 3.2718, 3.6752, 2.9440], device='cuda:1'), covar=tensor([0.0596, 0.1709, 0.1131, 0.2120, 0.0878, 0.0538, 0.0650, 0.1219], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0188, 0.0125, 0.0124, 0.0117, 0.0121, 0.0101, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 17:56:18,869 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8472, 1.5456, 1.8691, 2.0627, 1.3777, 1.7988, 1.9395, 1.8888], device='cuda:1'), covar=tensor([0.0057, 0.0105, 0.0056, 0.0034, 0.0120, 0.0125, 0.0056, 0.0042], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0226, 0.0344, 0.0280, 0.0224, 0.0274, 0.0245, 0.0260], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:56:49,479 INFO [train.py:873] (1/4) Epoch 8, batch 1600, loss[loss=0.1462, simple_loss=0.1489, pruned_loss=0.07171, over 2615.00 frames. ], tot_loss[loss=0.151, simple_loss=0.1727, pruned_loss=0.06468, over 1956711.49 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:56:51,887 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.201e+02 2.581e+02 3.079e+02 3.798e+02 6.375e+02, threshold=6.158e+02, percent-clipped=8.0 2022-12-07 17:57:27,847 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.83 vs. limit=5.0 2022-12-07 17:57:28,309 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1933, 2.1129, 4.2242, 2.7929, 4.0204, 2.0272, 3.1493, 3.9807], device='cuda:1'), covar=tensor([0.0461, 0.4725, 0.0298, 0.7702, 0.0437, 0.3559, 0.1277, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0237, 0.0180, 0.0317, 0.0204, 0.0239, 0.0228, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 17:57:36,811 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5411, 2.1177, 3.3934, 3.5203, 3.5837, 2.3001, 3.4681, 2.5365], device='cuda:1'), covar=tensor([0.0197, 0.0512, 0.0445, 0.0232, 0.0151, 0.0728, 0.0179, 0.0566], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0226, 0.0343, 0.0281, 0.0225, 0.0273, 0.0248, 0.0262], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 17:58:14,091 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 17:58:17,000 INFO [train.py:873] (1/4) Epoch 8, batch 1700, loss[loss=0.1849, simple_loss=0.1948, pruned_loss=0.08744, over 10327.00 frames. ], tot_loss[loss=0.1501, simple_loss=0.1722, pruned_loss=0.06396, over 1950231.13 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:58:19,791 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.375e+02 2.929e+02 3.612e+02 8.453e+02, threshold=5.858e+02, percent-clipped=6.0 2022-12-07 17:58:25,291 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-07 17:58:37,079 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2022-12-07 17:58:42,716 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:59:35,616 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9081, 1.3397, 2.1416, 1.1631, 2.0250, 2.1025, 1.8091, 2.1252], device='cuda:1'), covar=tensor([0.0250, 0.1519, 0.0258, 0.1633, 0.0341, 0.0364, 0.0601, 0.0270], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0157, 0.0149, 0.0167, 0.0161, 0.0163, 0.0132, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 17:59:36,606 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:59:44,435 INFO [train.py:873] (1/4) Epoch 8, batch 1800, loss[loss=0.164, simple_loss=0.1747, pruned_loss=0.07663, over 4991.00 frames. ], tot_loss[loss=0.151, simple_loss=0.1729, pruned_loss=0.06456, over 1939194.72 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:59:46,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.304e+02 2.792e+02 3.452e+02 5.919e+02, threshold=5.584e+02, percent-clipped=1.0 2022-12-07 18:00:27,890 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6603, 1.8626, 2.5374, 2.7103, 2.6677, 2.0211, 2.7608, 2.1611], device='cuda:1'), covar=tensor([0.0227, 0.0487, 0.0243, 0.0199, 0.0254, 0.0670, 0.0177, 0.0447], device='cuda:1'), in_proj_covar=tensor([0.0248, 0.0227, 0.0345, 0.0281, 0.0226, 0.0276, 0.0249, 0.0261], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:00:43,486 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2022-12-07 18:01:11,086 INFO [train.py:873] (1/4) Epoch 8, batch 1900, loss[loss=0.1566, simple_loss=0.1593, pruned_loss=0.07697, over 4943.00 frames. ], tot_loss[loss=0.1514, simple_loss=0.1734, pruned_loss=0.06468, over 2026393.71 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:01:13,578 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.876e+01 2.431e+02 3.063e+02 3.945e+02 8.424e+02, threshold=6.126e+02, percent-clipped=6.0 2022-12-07 18:01:28,887 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9269, 3.7262, 3.6579, 3.9693, 3.7490, 3.4684, 4.0107, 3.3088], device='cuda:1'), covar=tensor([0.0533, 0.0992, 0.0393, 0.0490, 0.0803, 0.1287, 0.0636, 0.0558], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0243, 0.0167, 0.0159, 0.0167, 0.0133, 0.0249, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 18:01:33,228 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4006, 2.3192, 1.7927, 1.8560, 2.2684, 2.3175, 2.3957, 2.3279], device='cuda:1'), covar=tensor([0.1417, 0.1003, 0.3615, 0.4615, 0.1765, 0.1608, 0.1838, 0.1590], device='cuda:1'), in_proj_covar=tensor([0.0331, 0.0243, 0.0403, 0.0510, 0.0298, 0.0375, 0.0365, 0.0315], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:01:35,133 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:01:35,884 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7824, 1.4059, 3.0786, 1.4076, 3.0955, 2.9750, 2.1046, 3.0846], device='cuda:1'), covar=tensor([0.0326, 0.2973, 0.0352, 0.2488, 0.0361, 0.0455, 0.0910, 0.0273], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0157, 0.0149, 0.0167, 0.0161, 0.0163, 0.0131, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:01:56,264 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4491, 1.3145, 1.3695, 0.7102, 1.4078, 0.6359, 1.1422, 1.7226], device='cuda:1'), covar=tensor([0.1214, 0.1807, 0.0706, 0.1994, 0.1765, 0.0948, 0.1386, 0.1494], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0022, 0.0020, 0.0021, 0.0030, 0.0021, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.9883e-05, 1.0074e-04, 1.0222e-04, 9.7288e-05, 1.0148e-04, 1.3175e-04, 1.0376e-04, 1.0440e-04], device='cuda:1') 2022-12-07 18:01:56,702 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-07 18:02:18,173 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 18:02:29,174 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:02:39,003 INFO [train.py:873] (1/4) Epoch 8, batch 2000, loss[loss=0.1607, simple_loss=0.1847, pruned_loss=0.06837, over 11174.00 frames. ], tot_loss[loss=0.1524, simple_loss=0.1739, pruned_loss=0.0655, over 1980897.12 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:02:41,554 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.941e+01 2.269e+02 2.935e+02 3.791e+02 7.498e+02, threshold=5.870e+02, percent-clipped=3.0 2022-12-07 18:02:47,389 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-07 18:03:03,133 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2825, 1.9101, 2.2917, 1.4786, 1.9289, 2.1662, 2.3092, 1.9834], device='cuda:1'), covar=tensor([0.0569, 0.0934, 0.0926, 0.1920, 0.1136, 0.0774, 0.0443, 0.1616], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0186, 0.0125, 0.0125, 0.0116, 0.0123, 0.0102, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 18:03:57,227 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:04:09,836 INFO [train.py:873] (1/4) Epoch 8, batch 2100, loss[loss=0.169, simple_loss=0.1903, pruned_loss=0.07389, over 14320.00 frames. ], tot_loss[loss=0.1507, simple_loss=0.1725, pruned_loss=0.06439, over 1987168.28 frames. ], batch size: 60, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:04:12,640 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.381e+02 2.883e+02 3.565e+02 6.243e+02, threshold=5.765e+02, percent-clipped=1.0 2022-12-07 18:04:24,759 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 18:05:31,153 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 18:05:37,625 INFO [train.py:873] (1/4) Epoch 8, batch 2200, loss[loss=0.2023, simple_loss=0.2041, pruned_loss=0.1002, over 8594.00 frames. ], tot_loss[loss=0.1518, simple_loss=0.1735, pruned_loss=0.06506, over 1920890.26 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:05:39,867 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.363e+02 3.011e+02 4.055e+02 7.336e+02, threshold=6.021e+02, percent-clipped=10.0 2022-12-07 18:05:48,523 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:05:59,194 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3258, 3.2752, 4.3297, 3.0157, 2.6748, 3.5508, 1.7828, 3.5347], device='cuda:1'), covar=tensor([0.0870, 0.1076, 0.0343, 0.2132, 0.2226, 0.0555, 0.4567, 0.0847], device='cuda:1'), in_proj_covar=tensor([0.0077, 0.0088, 0.0082, 0.0091, 0.0114, 0.0074, 0.0132, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:06:43,393 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:06:44,999 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:06:51,020 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:07:05,469 INFO [train.py:873] (1/4) Epoch 8, batch 2300, loss[loss=0.1581, simple_loss=0.1827, pruned_loss=0.06671, over 14287.00 frames. ], tot_loss[loss=0.1519, simple_loss=0.1733, pruned_loss=0.06522, over 1932560.52 frames. ], batch size: 44, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:07:08,045 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.336e+02 2.949e+02 3.747e+02 7.724e+02, threshold=5.898e+02, percent-clipped=2.0 2022-12-07 18:07:12,893 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2022-12-07 18:07:19,055 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9679, 2.6131, 4.9888, 3.3383, 4.7347, 2.3661, 3.8288, 4.5481], device='cuda:1'), covar=tensor([0.0374, 0.4458, 0.0369, 0.8017, 0.0434, 0.3629, 0.1006, 0.0270], device='cuda:1'), in_proj_covar=tensor([0.0229, 0.0235, 0.0182, 0.0316, 0.0204, 0.0237, 0.0223, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:07:39,187 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:07:42,696 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9436, 2.0237, 2.9807, 3.0604, 2.9815, 2.1724, 2.9225, 2.3202], device='cuda:1'), covar=tensor([0.0186, 0.0469, 0.0366, 0.0210, 0.0172, 0.0646, 0.0175, 0.0472], device='cuda:1'), in_proj_covar=tensor([0.0244, 0.0224, 0.0340, 0.0280, 0.0225, 0.0275, 0.0245, 0.0260], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:07:48,842 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 18:07:51,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0833, 4.6541, 4.6365, 5.0259, 4.7238, 4.3658, 5.0464, 4.3241], device='cuda:1'), covar=tensor([0.0272, 0.0905, 0.0287, 0.0427, 0.0641, 0.0499, 0.0539, 0.0455], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0247, 0.0167, 0.0158, 0.0165, 0.0132, 0.0248, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 18:08:21,819 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:08:24,726 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0495, 3.2713, 2.9591, 3.1062, 2.2820, 3.3225, 2.9021, 1.4205], device='cuda:1'), covar=tensor([0.3184, 0.0877, 0.1290, 0.0944, 0.1252, 0.0610, 0.1324, 0.3487], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0064, 0.0054, 0.0056, 0.0083, 0.0061, 0.0086, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 18:08:34,796 INFO [train.py:873] (1/4) Epoch 8, batch 2400, loss[loss=0.1191, simple_loss=0.1566, pruned_loss=0.04076, over 14642.00 frames. ], tot_loss[loss=0.1512, simple_loss=0.173, pruned_loss=0.0647, over 1908578.50 frames. ], batch size: 33, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:08:37,246 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.285e+02 2.947e+02 4.114e+02 7.749e+02, threshold=5.894e+02, percent-clipped=7.0 2022-12-07 18:09:04,219 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:09:14,531 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7611, 2.4490, 3.1677, 1.9904, 2.0388, 2.6580, 1.1973, 2.7398], device='cuda:1'), covar=tensor([0.1243, 0.1626, 0.0963, 0.3293, 0.3027, 0.1214, 0.6664, 0.1371], device='cuda:1'), in_proj_covar=tensor([0.0076, 0.0086, 0.0081, 0.0089, 0.0111, 0.0074, 0.0129, 0.0079], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:10:03,414 INFO [train.py:873] (1/4) Epoch 8, batch 2500, loss[loss=0.1415, simple_loss=0.1643, pruned_loss=0.05929, over 14231.00 frames. ], tot_loss[loss=0.149, simple_loss=0.1718, pruned_loss=0.06315, over 1951963.50 frames. ], batch size: 63, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:10:05,936 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 2.135e+02 2.789e+02 3.558e+02 8.555e+02, threshold=5.578e+02, percent-clipped=3.0 2022-12-07 18:10:50,330 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-07 18:11:02,497 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0401, 3.8672, 3.7724, 4.2191, 3.6924, 3.3986, 4.1936, 4.1487], device='cuda:1'), covar=tensor([0.0724, 0.0666, 0.0759, 0.0523, 0.0792, 0.0769, 0.0519, 0.0688], device='cuda:1'), in_proj_covar=tensor([0.0119, 0.0108, 0.0121, 0.0124, 0.0126, 0.0097, 0.0137, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 18:11:03,342 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:11:15,549 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:11:22,973 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6673, 5.3727, 4.8944, 5.1927, 5.0216, 5.4342, 5.5852, 5.5602], device='cuda:1'), covar=tensor([0.0525, 0.0314, 0.1653, 0.2383, 0.0674, 0.0653, 0.0676, 0.0741], device='cuda:1'), in_proj_covar=tensor([0.0336, 0.0244, 0.0399, 0.0514, 0.0300, 0.0376, 0.0368, 0.0322], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:11:30,375 INFO [train.py:873] (1/4) Epoch 8, batch 2600, loss[loss=0.1232, simple_loss=0.1263, pruned_loss=0.06001, over 1311.00 frames. ], tot_loss[loss=0.1497, simple_loss=0.1721, pruned_loss=0.06363, over 2010302.30 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:11:32,918 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.392e+02 2.780e+02 3.615e+02 5.645e+02, threshold=5.559e+02, percent-clipped=1.0 2022-12-07 18:11:57,679 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:11:58,676 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:12:57,534 INFO [train.py:873] (1/4) Epoch 8, batch 2700, loss[loss=0.1502, simple_loss=0.1747, pruned_loss=0.06286, over 14270.00 frames. ], tot_loss[loss=0.1505, simple_loss=0.1723, pruned_loss=0.06437, over 1981131.61 frames. ], batch size: 76, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:13:00,127 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.995e+01 2.381e+02 3.010e+02 3.860e+02 1.123e+03, threshold=6.020e+02, percent-clipped=6.0 2022-12-07 18:13:23,001 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0137, 3.7927, 3.7163, 4.0816, 3.6227, 3.4096, 4.0785, 4.0259], device='cuda:1'), covar=tensor([0.0681, 0.0803, 0.0716, 0.0546, 0.0831, 0.0718, 0.0590, 0.0618], device='cuda:1'), in_proj_covar=tensor([0.0123, 0.0114, 0.0125, 0.0130, 0.0130, 0.0101, 0.0142, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-07 18:13:37,288 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8148, 1.3132, 2.6712, 2.5248, 2.6650, 2.6881, 2.2031, 2.7281], device='cuda:1'), covar=tensor([0.1002, 0.1248, 0.0151, 0.0297, 0.0265, 0.0153, 0.0394, 0.0187], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0154, 0.0115, 0.0158, 0.0132, 0.0129, 0.0108, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:13:40,321 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0999, 1.9016, 2.2488, 2.2002, 1.9723, 1.8694, 2.2533, 1.9079], device='cuda:1'), covar=tensor([0.0126, 0.0298, 0.0115, 0.0093, 0.0188, 0.0370, 0.0164, 0.0170], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0226, 0.0346, 0.0285, 0.0228, 0.0274, 0.0246, 0.0262], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:14:09,291 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2054, 2.1257, 2.4353, 1.6358, 1.7078, 2.3024, 1.2313, 2.0755], device='cuda:1'), covar=tensor([0.0811, 0.1352, 0.0734, 0.2083, 0.2707, 0.0649, 0.4207, 0.0857], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0087, 0.0081, 0.0088, 0.0112, 0.0074, 0.0128, 0.0079], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:14:27,408 INFO [train.py:873] (1/4) Epoch 8, batch 2800, loss[loss=0.1698, simple_loss=0.1628, pruned_loss=0.08837, over 2640.00 frames. ], tot_loss[loss=0.1505, simple_loss=0.1725, pruned_loss=0.06424, over 1966912.22 frames. ], batch size: 100, lr: 9.99e-03, grad_scale: 8.0 2022-12-07 18:14:27,641 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5477, 1.7221, 1.2535, 1.6989, 1.5803, 1.3060, 1.4928, 1.4815], device='cuda:1'), covar=tensor([0.0631, 0.0849, 0.1244, 0.1279, 0.1197, 0.0494, 0.0608, 0.0674], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0014, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.2874e-05, 9.0478e-05, 7.9668e-05, 8.3110e-05, 8.4022e-05, 1.2162e-04, 1.0198e-04, 1.1736e-04], device='cuda:1') 2022-12-07 18:14:30,874 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.464e+02 2.891e+02 3.921e+02 7.148e+02, threshold=5.781e+02, percent-clipped=1.0 2022-12-07 18:14:38,386 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:28,248 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:32,978 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:50,414 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0196, 2.0515, 4.6041, 4.2831, 4.1358, 4.7908, 4.4603, 4.7227], device='cuda:1'), covar=tensor([0.1281, 0.1194, 0.0101, 0.0164, 0.0157, 0.0099, 0.0086, 0.0121], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0153, 0.0114, 0.0156, 0.0131, 0.0127, 0.0107, 0.0112], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:15:54,463 INFO [train.py:873] (1/4) Epoch 8, batch 2900, loss[loss=0.1647, simple_loss=0.1825, pruned_loss=0.07345, over 12791.00 frames. ], tot_loss[loss=0.1515, simple_loss=0.1733, pruned_loss=0.06486, over 1973705.36 frames. ], batch size: 100, lr: 9.99e-03, grad_scale: 8.0 2022-12-07 18:15:57,949 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.733e+01 2.419e+02 2.995e+02 3.748e+02 5.974e+02, threshold=5.989e+02, percent-clipped=2.0 2022-12-07 18:16:06,715 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:09,830 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:12,514 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1235, 1.2423, 1.2157, 1.7332, 1.0256, 1.0466, 1.4266, 1.3163], device='cuda:1'), covar=tensor([0.2188, 0.1783, 0.0965, 0.0911, 0.1639, 0.0792, 0.0718, 0.1044], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.4151e-05, 9.2082e-05, 8.1814e-05, 8.5112e-05, 8.5525e-05, 1.2367e-04, 1.0426e-04, 1.1958e-04], device='cuda:1') 2022-12-07 18:16:22,847 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:28,089 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-07 18:16:37,749 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8832, 2.9092, 2.9270, 2.8506, 2.2664, 3.0142, 2.7680, 1.3115], device='cuda:1'), covar=tensor([0.2999, 0.1070, 0.1407, 0.1166, 0.1353, 0.0862, 0.2021, 0.3951], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0066, 0.0055, 0.0057, 0.0083, 0.0063, 0.0088, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 18:16:57,694 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3184, 2.9040, 2.8927, 2.0018, 2.6775, 2.9874, 3.1647, 2.5291], device='cuda:1'), covar=tensor([0.0546, 0.1328, 0.1146, 0.2180, 0.1012, 0.0573, 0.0924, 0.1530], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0187, 0.0126, 0.0125, 0.0117, 0.0122, 0.0104, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 18:16:59,454 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:59,904 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 18:17:04,332 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:17:22,325 INFO [train.py:873] (1/4) Epoch 8, batch 3000, loss[loss=0.1538, simple_loss=0.1818, pruned_loss=0.06287, over 14259.00 frames. ], tot_loss[loss=0.1515, simple_loss=0.1734, pruned_loss=0.06475, over 1967830.15 frames. ], batch size: 44, lr: 9.98e-03, grad_scale: 8.0 2022-12-07 18:17:22,326 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 18:17:42,271 INFO [train.py:905] (1/4) Epoch 8, validation: loss=0.1226, simple_loss=0.1659, pruned_loss=0.03968, over 857387.00 frames. 2022-12-07 18:17:42,272 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 18:17:45,688 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 2.436e+02 3.032e+02 3.912e+02 1.128e+03, threshold=6.064e+02, percent-clipped=7.0 2022-12-07 18:18:06,882 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2955, 4.0335, 3.7582, 3.8898, 4.1583, 4.1841, 4.3017, 4.2684], device='cuda:1'), covar=tensor([0.0949, 0.0568, 0.2123, 0.3197, 0.0689, 0.0873, 0.1074, 0.0902], device='cuda:1'), in_proj_covar=tensor([0.0339, 0.0246, 0.0403, 0.0513, 0.0292, 0.0374, 0.0370, 0.0324], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:18:46,445 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 18:19:11,700 INFO [train.py:873] (1/4) Epoch 8, batch 3100, loss[loss=0.153, simple_loss=0.177, pruned_loss=0.06446, over 14661.00 frames. ], tot_loss[loss=0.15, simple_loss=0.1722, pruned_loss=0.06386, over 2001417.90 frames. ], batch size: 33, lr: 9.97e-03, grad_scale: 8.0 2022-12-07 18:19:15,078 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.402e+01 2.241e+02 2.784e+02 3.482e+02 7.776e+02, threshold=5.568e+02, percent-clipped=3.0 2022-12-07 18:19:47,660 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6090, 2.3584, 2.4194, 1.5618, 2.2293, 2.4014, 2.7167, 2.2237], device='cuda:1'), covar=tensor([0.0713, 0.1436, 0.1044, 0.2204, 0.1097, 0.0627, 0.0592, 0.1586], device='cuda:1'), in_proj_covar=tensor([0.0121, 0.0185, 0.0124, 0.0123, 0.0115, 0.0119, 0.0103, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 18:19:57,932 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6808, 1.3853, 2.9325, 1.3847, 3.0036, 2.9353, 2.2069, 3.0565], device='cuda:1'), covar=tensor([0.0269, 0.2513, 0.0307, 0.2077, 0.0274, 0.0348, 0.0767, 0.0189], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0160, 0.0151, 0.0168, 0.0164, 0.0164, 0.0131, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:20:01,991 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7170, 1.4290, 2.9293, 1.5103, 3.0152, 2.9504, 2.1557, 3.0502], device='cuda:1'), covar=tensor([0.0282, 0.2532, 0.0309, 0.1908, 0.0272, 0.0348, 0.0825, 0.0197], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0159, 0.0151, 0.0168, 0.0163, 0.0163, 0.0130, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:20:11,544 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:20:38,655 INFO [train.py:873] (1/4) Epoch 8, batch 3200, loss[loss=0.1333, simple_loss=0.1584, pruned_loss=0.05409, over 13907.00 frames. ], tot_loss[loss=0.1506, simple_loss=0.1725, pruned_loss=0.06431, over 1999546.07 frames. ], batch size: 20, lr: 9.96e-03, grad_scale: 8.0 2022-12-07 18:20:42,716 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.292e+02 3.048e+02 3.641e+02 9.234e+02, threshold=6.095e+02, percent-clipped=4.0 2022-12-07 18:20:47,426 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:21:08,711 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2022-12-07 18:21:40,189 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:21:42,063 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:21:49,853 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9480, 1.9854, 1.6252, 2.0660, 1.8412, 1.9227, 1.8489, 1.7631], device='cuda:1'), covar=tensor([0.0725, 0.0836, 0.1579, 0.0391, 0.0852, 0.0523, 0.1287, 0.0793], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0296, 0.0275, 0.0218, 0.0279, 0.0274, 0.0257, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:22:02,111 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8569, 2.1706, 2.7113, 2.3201, 2.8556, 2.5582, 2.7214, 2.3801], device='cuda:1'), covar=tensor([0.0621, 0.2878, 0.0950, 0.1775, 0.0785, 0.0781, 0.1170, 0.1833], device='cuda:1'), in_proj_covar=tensor([0.0316, 0.0319, 0.0384, 0.0298, 0.0363, 0.0299, 0.0354, 0.0324], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:22:08,269 INFO [train.py:873] (1/4) Epoch 8, batch 3300, loss[loss=0.1578, simple_loss=0.1761, pruned_loss=0.06971, over 14214.00 frames. ], tot_loss[loss=0.1507, simple_loss=0.1723, pruned_loss=0.06457, over 1916890.25 frames. ], batch size: 94, lr: 9.95e-03, grad_scale: 8.0 2022-12-07 18:22:12,320 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.398e+02 3.040e+02 3.643e+02 7.318e+02, threshold=6.080e+02, percent-clipped=1.0 2022-12-07 18:22:15,575 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 18:22:57,363 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7978, 0.9142, 0.9139, 0.8105, 0.8347, 0.4382, 0.6513, 0.5785], device='cuda:1'), covar=tensor([0.0251, 0.0198, 0.0224, 0.0174, 0.0312, 0.0626, 0.0365, 0.0619], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0012, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.5572e-05, 9.3100e-05, 8.3842e-05, 8.7440e-05, 8.5105e-05, 1.2704e-04, 1.0603e-04, 1.2192e-04], device='cuda:1') 2022-12-07 18:23:32,870 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3396, 1.5664, 1.6026, 1.4682, 1.7454, 1.0600, 1.1145, 1.0184], device='cuda:1'), covar=tensor([0.0608, 0.0814, 0.0526, 0.1190, 0.0394, 0.0367, 0.0303, 0.0728], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.5942e-05, 9.2917e-05, 8.3721e-05, 8.7707e-05, 8.5024e-05, 1.2693e-04, 1.0598e-04, 1.2184e-04], device='cuda:1') 2022-12-07 18:23:34,512 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7739, 0.8662, 0.6056, 0.7952, 0.8160, 0.2930, 0.6571, 0.6630], device='cuda:1'), covar=tensor([0.0244, 0.0265, 0.0251, 0.0185, 0.0184, 0.0199, 0.0972, 0.0442], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0021, 0.0023, 0.0021, 0.0022, 0.0031, 0.0023, 0.0022], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 18:23:35,253 INFO [train.py:873] (1/4) Epoch 8, batch 3400, loss[loss=0.2101, simple_loss=0.1805, pruned_loss=0.1198, over 1168.00 frames. ], tot_loss[loss=0.1513, simple_loss=0.1727, pruned_loss=0.0649, over 1942723.97 frames. ], batch size: 100, lr: 9.94e-03, grad_scale: 8.0 2022-12-07 18:23:37,910 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7186, 2.3623, 3.2362, 2.2690, 2.1127, 2.8547, 1.3541, 2.7305], device='cuda:1'), covar=tensor([0.1174, 0.2028, 0.0738, 0.2698, 0.3001, 0.1025, 0.5844, 0.1311], device='cuda:1'), in_proj_covar=tensor([0.0076, 0.0087, 0.0081, 0.0090, 0.0110, 0.0076, 0.0129, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:23:39,443 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.408e+02 2.894e+02 3.622e+02 6.016e+02, threshold=5.789e+02, percent-clipped=0.0 2022-12-07 18:23:57,909 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-07 18:24:36,620 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:25:03,323 INFO [train.py:873] (1/4) Epoch 8, batch 3500, loss[loss=0.1346, simple_loss=0.1672, pruned_loss=0.05104, over 14268.00 frames. ], tot_loss[loss=0.1516, simple_loss=0.1728, pruned_loss=0.0652, over 1944482.35 frames. ], batch size: 57, lr: 9.93e-03, grad_scale: 8.0 2022-12-07 18:25:07,749 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.219e+02 2.782e+02 3.549e+02 6.206e+02, threshold=5.564e+02, percent-clipped=1.0 2022-12-07 18:25:18,096 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:25:50,545 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2022-12-07 18:26:01,182 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:03,911 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:20,795 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:26:30,711 INFO [train.py:873] (1/4) Epoch 8, batch 3600, loss[loss=0.1424, simple_loss=0.1625, pruned_loss=0.0612, over 14170.00 frames. ], tot_loss[loss=0.1507, simple_loss=0.1722, pruned_loss=0.06454, over 1915799.75 frames. ], batch size: 37, lr: 9.92e-03, grad_scale: 8.0 2022-12-07 18:26:35,120 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.906e+01 2.280e+02 2.872e+02 3.796e+02 9.491e+02, threshold=5.743e+02, percent-clipped=4.0 2022-12-07 18:26:46,158 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:49,004 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0543, 2.7187, 3.7344, 2.9730, 3.8939, 3.6922, 3.6895, 3.0813], device='cuda:1'), covar=tensor([0.0523, 0.2714, 0.1042, 0.1797, 0.0852, 0.0749, 0.1593, 0.1950], device='cuda:1'), in_proj_covar=tensor([0.0316, 0.0317, 0.0384, 0.0301, 0.0367, 0.0302, 0.0355, 0.0325], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:27:14,178 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 18:27:20,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9312, 3.2694, 4.5523, 3.4962, 4.6384, 4.4982, 4.3694, 3.9111], device='cuda:1'), covar=tensor([0.0402, 0.2815, 0.0736, 0.1666, 0.0669, 0.0688, 0.1605, 0.2083], device='cuda:1'), in_proj_covar=tensor([0.0316, 0.0318, 0.0384, 0.0301, 0.0367, 0.0303, 0.0355, 0.0325], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:27:39,513 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8723, 1.6227, 4.7758, 4.2915, 4.3220, 4.9090, 4.6698, 4.8690], device='cuda:1'), covar=tensor([0.1494, 0.1608, 0.0124, 0.0189, 0.0179, 0.0164, 0.0092, 0.0181], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0154, 0.0115, 0.0159, 0.0135, 0.0129, 0.0108, 0.0113], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:27:45,019 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2022-12-07 18:27:59,551 INFO [train.py:873] (1/4) Epoch 8, batch 3700, loss[loss=0.1873, simple_loss=0.1571, pruned_loss=0.1087, over 1224.00 frames. ], tot_loss[loss=0.1498, simple_loss=0.1718, pruned_loss=0.06393, over 1918909.93 frames. ], batch size: 100, lr: 9.92e-03, grad_scale: 8.0 2022-12-07 18:28:00,539 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9191, 1.3389, 2.0718, 1.3350, 2.0461, 2.0885, 1.8066, 2.1251], device='cuda:1'), covar=tensor([0.0276, 0.1627, 0.0316, 0.1432, 0.0382, 0.0387, 0.0710, 0.0301], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0160, 0.0152, 0.0168, 0.0166, 0.0166, 0.0132, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:28:02,496 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7846, 2.0685, 2.8204, 2.7881, 2.7550, 1.9775, 2.8574, 2.1681], device='cuda:1'), covar=tensor([0.0206, 0.0500, 0.0357, 0.0234, 0.0237, 0.0723, 0.0179, 0.0555], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0228, 0.0339, 0.0283, 0.0228, 0.0275, 0.0249, 0.0261], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:28:03,912 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.170e+02 2.783e+02 3.492e+02 8.002e+02, threshold=5.566e+02, percent-clipped=4.0 2022-12-07 18:28:11,730 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5888, 3.3921, 3.2088, 2.2362, 3.0877, 3.3345, 3.4810, 2.9329], device='cuda:1'), covar=tensor([0.0480, 0.1216, 0.0818, 0.1799, 0.0779, 0.0535, 0.0859, 0.1231], device='cuda:1'), in_proj_covar=tensor([0.0122, 0.0189, 0.0125, 0.0125, 0.0117, 0.0124, 0.0104, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 18:29:25,509 INFO [train.py:873] (1/4) Epoch 8, batch 3800, loss[loss=0.1713, simple_loss=0.1619, pruned_loss=0.0903, over 3883.00 frames. ], tot_loss[loss=0.1508, simple_loss=0.1725, pruned_loss=0.06458, over 1878576.99 frames. ], batch size: 100, lr: 9.91e-03, grad_scale: 8.0 2022-12-07 18:29:29,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.607e+02 3.160e+02 3.908e+02 8.359e+02, threshold=6.320e+02, percent-clipped=5.0 2022-12-07 18:30:18,210 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4484, 3.1641, 3.1431, 3.4363, 3.2720, 3.4231, 3.5265, 2.8332], device='cuda:1'), covar=tensor([0.0445, 0.1141, 0.0486, 0.0526, 0.0848, 0.0366, 0.0590, 0.0667], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0245, 0.0166, 0.0160, 0.0163, 0.0130, 0.0243, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 18:30:24,813 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:30:26,493 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:30:54,312 INFO [train.py:873] (1/4) Epoch 8, batch 3900, loss[loss=0.1353, simple_loss=0.1674, pruned_loss=0.05158, over 14642.00 frames. ], tot_loss[loss=0.1496, simple_loss=0.172, pruned_loss=0.06363, over 1935657.20 frames. ], batch size: 33, lr: 9.90e-03, grad_scale: 8.0 2022-12-07 18:30:58,940 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.187e+02 2.677e+02 3.401e+02 6.261e+02, threshold=5.355e+02, percent-clipped=0.0 2022-12-07 18:31:04,774 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8173, 2.4989, 3.5419, 2.3808, 2.2877, 2.9647, 1.6001, 2.7067], device='cuda:1'), covar=tensor([0.1335, 0.1251, 0.0398, 0.2070, 0.2347, 0.0720, 0.4379, 0.1216], device='cuda:1'), in_proj_covar=tensor([0.0073, 0.0084, 0.0078, 0.0087, 0.0108, 0.0073, 0.0123, 0.0078], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:31:06,452 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:31:17,993 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3119, 1.5436, 1.4064, 1.3356, 1.4986, 1.0135, 0.9204, 0.8450], device='cuda:1'), covar=tensor([0.0396, 0.1004, 0.0658, 0.0487, 0.0877, 0.0474, 0.0332, 0.0972], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.7697e-05, 9.5417e-05, 8.4227e-05, 8.8837e-05, 8.6499e-05, 1.3101e-04, 1.0872e-04, 1.2368e-04], device='cuda:1') 2022-12-07 18:31:19,849 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:31:20,041 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.40 vs. limit=5.0 2022-12-07 18:31:33,516 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:31:58,962 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:32:03,909 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2837, 2.3205, 3.3045, 3.4678, 3.3200, 2.2900, 3.4538, 2.6934], device='cuda:1'), covar=tensor([0.0206, 0.0459, 0.0472, 0.0231, 0.0191, 0.0677, 0.0194, 0.0470], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0224, 0.0339, 0.0280, 0.0225, 0.0270, 0.0247, 0.0259], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:32:20,971 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-07 18:32:22,732 INFO [train.py:873] (1/4) Epoch 8, batch 4000, loss[loss=0.1713, simple_loss=0.186, pruned_loss=0.07829, over 11989.00 frames. ], tot_loss[loss=0.1496, simple_loss=0.1719, pruned_loss=0.06363, over 1945959.70 frames. ], batch size: 100, lr: 9.89e-03, grad_scale: 8.0 2022-12-07 18:32:27,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 2.242e+02 2.985e+02 3.676e+02 7.027e+02, threshold=5.970e+02, percent-clipped=5.0 2022-12-07 18:32:53,076 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:02,983 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 18:33:14,748 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:16,334 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:32,580 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 18:33:51,678 INFO [train.py:873] (1/4) Epoch 8, batch 4100, loss[loss=0.2016, simple_loss=0.1673, pruned_loss=0.118, over 1281.00 frames. ], tot_loss[loss=0.1502, simple_loss=0.1721, pruned_loss=0.06414, over 1892588.95 frames. ], batch size: 100, lr: 9.88e-03, grad_scale: 8.0 2022-12-07 18:33:56,137 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 2.377e+02 3.103e+02 3.855e+02 1.239e+03, threshold=6.205e+02, percent-clipped=7.0 2022-12-07 18:34:02,712 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2497, 1.1550, 1.4594, 1.1906, 1.3420, 0.7076, 1.1728, 1.1050], device='cuda:1'), covar=tensor([0.0814, 0.1486, 0.0773, 0.0966, 0.0826, 0.0723, 0.0602, 0.1429], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0012, 0.0012, 0.0019, 0.0015, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.8527e-05, 9.6227e-05, 8.5255e-05, 9.0786e-05, 8.7200e-05, 1.3128e-04, 1.0948e-04, 1.2504e-04], device='cuda:1') 2022-12-07 18:34:08,947 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:34:09,301 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2022-12-07 18:34:10,564 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:34:48,077 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4218, 1.3322, 1.5893, 1.6578, 1.4338, 0.9739, 1.1243, 1.1576], device='cuda:1'), covar=tensor([0.0522, 0.1258, 0.0584, 0.0386, 0.0827, 0.0489, 0.0449, 0.0896], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.7589e-05, 9.4585e-05, 8.4342e-05, 8.9160e-05, 8.5844e-05, 1.2925e-04, 1.0911e-04, 1.2331e-04], device='cuda:1') 2022-12-07 18:35:20,884 INFO [train.py:873] (1/4) Epoch 8, batch 4200, loss[loss=0.1157, simple_loss=0.1162, pruned_loss=0.05755, over 1274.00 frames. ], tot_loss[loss=0.1508, simple_loss=0.1723, pruned_loss=0.06468, over 1918265.79 frames. ], batch size: 100, lr: 9.87e-03, grad_scale: 8.0 2022-12-07 18:35:25,040 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.260e+02 2.916e+02 3.546e+02 6.128e+02, threshold=5.833e+02, percent-clipped=0.0 2022-12-07 18:35:42,103 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:35:59,643 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:36:06,013 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:15,603 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:32,009 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2022-12-07 18:36:41,991 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:36:48,819 INFO [train.py:873] (1/4) Epoch 8, batch 4300, loss[loss=0.1583, simple_loss=0.1774, pruned_loss=0.06965, over 10376.00 frames. ], tot_loss[loss=0.1502, simple_loss=0.1721, pruned_loss=0.06417, over 1933364.08 frames. ], batch size: 100, lr: 9.86e-03, grad_scale: 8.0 2022-12-07 18:36:53,098 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 2.274e+02 2.646e+02 3.498e+02 8.636e+02, threshold=5.293e+02, percent-clipped=3.0 2022-12-07 18:36:57,018 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:59,594 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:37:08,955 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:37:14,073 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:37:49,426 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:15,734 INFO [train.py:873] (1/4) Epoch 8, batch 4400, loss[loss=0.1432, simple_loss=0.1763, pruned_loss=0.055, over 14477.00 frames. ], tot_loss[loss=0.1498, simple_loss=0.1722, pruned_loss=0.06367, over 2012999.27 frames. ], batch size: 49, lr: 9.86e-03, grad_scale: 8.0 2022-12-07 18:38:19,656 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.383e+01 2.226e+02 2.824e+02 3.594e+02 6.936e+02, threshold=5.649e+02, percent-clipped=3.0 2022-12-07 18:38:27,791 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:29,518 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:35,719 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1852, 1.4271, 1.3578, 1.2613, 1.2355, 0.8400, 0.9202, 0.8251], device='cuda:1'), covar=tensor([0.0179, 0.0410, 0.0180, 0.0177, 0.0244, 0.0392, 0.0250, 0.0584], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.5757e-05, 9.2222e-05, 8.3740e-05, 8.7480e-05, 8.5708e-05, 1.2726e-04, 1.0669e-04, 1.2179e-04], device='cuda:1') 2022-12-07 18:39:10,886 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:39:41,406 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-07 18:39:43,841 INFO [train.py:873] (1/4) Epoch 8, batch 4500, loss[loss=0.1522, simple_loss=0.1666, pruned_loss=0.06893, over 5970.00 frames. ], tot_loss[loss=0.148, simple_loss=0.171, pruned_loss=0.06251, over 1972689.82 frames. ], batch size: 100, lr: 9.85e-03, grad_scale: 8.0 2022-12-07 18:39:47,866 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.294e+02 2.953e+02 3.643e+02 6.271e+02, threshold=5.907e+02, percent-clipped=3.0 2022-12-07 18:40:04,373 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:04,436 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:07,033 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:07,972 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5054, 3.0899, 2.4558, 3.6090, 3.4098, 3.4964, 2.9630, 2.4674], device='cuda:1'), covar=tensor([0.0804, 0.1676, 0.3999, 0.0476, 0.0866, 0.1169, 0.1424, 0.4256], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0303, 0.0280, 0.0225, 0.0286, 0.0282, 0.0258, 0.0270], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 18:40:09,032 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-07 18:40:35,491 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:45,757 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:00,044 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:10,289 INFO [train.py:873] (1/4) Epoch 8, batch 4600, loss[loss=0.1313, simple_loss=0.1518, pruned_loss=0.05538, over 4986.00 frames. ], tot_loss[loss=0.149, simple_loss=0.1718, pruned_loss=0.06307, over 2006633.11 frames. ], batch size: 100, lr: 9.84e-03, grad_scale: 8.0 2022-12-07 18:41:14,686 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.523e+02 3.027e+02 3.692e+02 7.181e+02, threshold=6.055e+02, percent-clipped=3.0 2022-12-07 18:41:16,384 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:26,197 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 18:41:29,246 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:35,984 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:56,006 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7513, 3.5055, 3.1549, 2.5772, 3.1752, 3.4977, 3.7546, 2.7269], device='cuda:1'), covar=tensor([0.0819, 0.2291, 0.1357, 0.2210, 0.0976, 0.0705, 0.0841, 0.1948], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0190, 0.0128, 0.0127, 0.0120, 0.0127, 0.0106, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:1') 2022-12-07 18:42:07,622 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:18,288 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:38,954 INFO [train.py:873] (1/4) Epoch 8, batch 4700, loss[loss=0.1645, simple_loss=0.1765, pruned_loss=0.07631, over 7690.00 frames. ], tot_loss[loss=0.1496, simple_loss=0.1719, pruned_loss=0.06361, over 1952840.42 frames. ], batch size: 100, lr: 9.83e-03, grad_scale: 8.0 2022-12-07 18:42:43,279 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.451e+02 2.895e+02 3.498e+02 7.656e+02, threshold=5.791e+02, percent-clipped=2.0 2022-12-07 18:42:51,307 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:52,959 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:27,661 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7123, 2.9320, 4.3072, 3.5879, 4.4012, 4.4075, 4.1485, 3.9242], device='cuda:1'), covar=tensor([0.0503, 0.2785, 0.0815, 0.1572, 0.0721, 0.0568, 0.1679, 0.1688], device='cuda:1'), in_proj_covar=tensor([0.0314, 0.0315, 0.0379, 0.0304, 0.0365, 0.0297, 0.0348, 0.0323], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:43:33,484 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:35,288 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:35,469 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3321, 1.8679, 1.8525, 1.5254, 1.3606, 1.5491, 1.7493, 1.4726], device='cuda:1'), covar=tensor([0.1035, 0.1028, 0.0736, 0.0802, 0.0823, 0.0389, 0.0355, 0.0985], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0018, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.5008e-05, 9.0603e-05, 8.1952e-05, 8.6887e-05, 8.4518e-05, 1.2611e-04, 1.0594e-04, 1.2056e-04], device='cuda:1') 2022-12-07 18:44:05,078 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:44:06,679 INFO [train.py:873] (1/4) Epoch 8, batch 4800, loss[loss=0.1567, simple_loss=0.179, pruned_loss=0.06718, over 14256.00 frames. ], tot_loss[loss=0.1496, simple_loss=0.1721, pruned_loss=0.06351, over 1964807.41 frames. ], batch size: 76, lr: 9.82e-03, grad_scale: 8.0 2022-12-07 18:44:10,752 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.410e+02 2.943e+02 4.035e+02 9.695e+02, threshold=5.886e+02, percent-clipped=7.0 2022-12-07 18:44:23,493 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:44:47,115 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5727, 1.3755, 4.1516, 2.1581, 4.1715, 4.3647, 3.7534, 4.9032], device='cuda:1'), covar=tensor([0.0170, 0.3184, 0.0359, 0.2126, 0.0299, 0.0330, 0.0475, 0.0109], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0158, 0.0151, 0.0168, 0.0164, 0.0165, 0.0132, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 18:44:58,951 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:20,565 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:36,247 INFO [train.py:873] (1/4) Epoch 8, batch 4900, loss[loss=0.1743, simple_loss=0.1854, pruned_loss=0.08158, over 8583.00 frames. ], tot_loss[loss=0.1497, simple_loss=0.172, pruned_loss=0.06372, over 1873942.64 frames. ], batch size: 100, lr: 9.81e-03, grad_scale: 8.0 2022-12-07 18:45:40,366 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.301e+02 2.954e+02 3.459e+02 6.054e+02, threshold=5.907e+02, percent-clipped=1.0 2022-12-07 18:45:42,382 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:50,037 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:51,644 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:46:24,223 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:33,365 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:34,131 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:40,138 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6493, 1.1504, 1.3169, 1.2885, 1.1201, 1.3656, 1.1209, 0.8269], device='cuda:1'), covar=tensor([0.2250, 0.0772, 0.0324, 0.0368, 0.1175, 0.0348, 0.1275, 0.1391], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0067, 0.0056, 0.0057, 0.0087, 0.0064, 0.0087, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 18:46:41,136 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7216, 2.2228, 3.7588, 3.8148, 3.7495, 2.2989, 3.8179, 2.8140], device='cuda:1'), covar=tensor([0.0259, 0.0608, 0.0612, 0.0308, 0.0200, 0.0942, 0.0223, 0.0645], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0229, 0.0342, 0.0289, 0.0229, 0.0275, 0.0249, 0.0262], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:46:55,827 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 18:47:03,918 INFO [train.py:873] (1/4) Epoch 8, batch 5000, loss[loss=0.1286, simple_loss=0.1676, pruned_loss=0.04474, over 14317.00 frames. ], tot_loss[loss=0.148, simple_loss=0.1713, pruned_loss=0.06232, over 1946001.08 frames. ], batch size: 28, lr: 9.80e-03, grad_scale: 8.0 2022-12-07 18:47:08,078 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.289e+02 3.008e+02 3.693e+02 7.100e+02, threshold=6.016e+02, percent-clipped=4.0 2022-12-07 18:47:15,005 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:48:31,876 INFO [train.py:873] (1/4) Epoch 8, batch 5100, loss[loss=0.1266, simple_loss=0.162, pruned_loss=0.04567, over 14243.00 frames. ], tot_loss[loss=0.148, simple_loss=0.1711, pruned_loss=0.06245, over 1943045.99 frames. ], batch size: 35, lr: 9.80e-03, grad_scale: 8.0 2022-12-07 18:48:36,395 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.346e+02 3.025e+02 3.888e+02 7.693e+02, threshold=6.049e+02, percent-clipped=2.0 2022-12-07 18:48:48,773 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:19,568 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:26,469 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0960, 2.1263, 1.9816, 2.2015, 1.7587, 1.9391, 2.1218, 2.0963], device='cuda:1'), covar=tensor([0.0859, 0.0992, 0.1090, 0.0813, 0.1305, 0.0791, 0.0992, 0.0921], device='cuda:1'), in_proj_covar=tensor([0.0124, 0.0117, 0.0125, 0.0131, 0.0129, 0.0101, 0.0144, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 18:49:30,697 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:45,335 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:59,577 INFO [train.py:873] (1/4) Epoch 8, batch 5200, loss[loss=0.1327, simple_loss=0.167, pruned_loss=0.04922, over 13554.00 frames. ], tot_loss[loss=0.1489, simple_loss=0.1716, pruned_loss=0.06305, over 1949486.87 frames. ], batch size: 100, lr: 9.79e-03, grad_scale: 16.0 2022-12-07 18:50:00,111 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6594, 1.9820, 2.6828, 2.6784, 2.6105, 2.0579, 2.7184, 2.1118], device='cuda:1'), covar=tensor([0.0217, 0.0553, 0.0334, 0.0236, 0.0251, 0.0706, 0.0185, 0.0477], device='cuda:1'), in_proj_covar=tensor([0.0248, 0.0228, 0.0341, 0.0287, 0.0228, 0.0274, 0.0246, 0.0259], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 18:50:04,075 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 2.199e+02 2.811e+02 3.642e+02 7.366e+02, threshold=5.622e+02, percent-clipped=3.0 2022-12-07 18:50:05,164 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:07,747 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:13,742 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:26,669 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:31,977 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9596, 2.0652, 4.0073, 2.7277, 3.8398, 1.9535, 3.0683, 3.7863], device='cuda:1'), covar=tensor([0.0574, 0.5273, 0.0520, 0.7737, 0.0467, 0.4301, 0.1289, 0.0407], device='cuda:1'), in_proj_covar=tensor([0.0235, 0.0233, 0.0182, 0.0313, 0.0201, 0.0234, 0.0224, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:50:40,814 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4881, 1.2186, 1.4858, 1.1032, 1.5510, 0.5807, 1.6221, 1.7739], device='cuda:1'), covar=tensor([0.1821, 0.1514, 0.1182, 0.1706, 0.1978, 0.0822, 0.1420, 0.1088], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0023, 0.0023, 0.0022, 0.0023, 0.0032, 0.0023, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 18:50:56,060 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:59,068 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:51:01,505 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:51:27,323 INFO [train.py:873] (1/4) Epoch 8, batch 5300, loss[loss=0.1609, simple_loss=0.1818, pruned_loss=0.07001, over 11958.00 frames. ], tot_loss[loss=0.1496, simple_loss=0.1722, pruned_loss=0.06352, over 1908114.85 frames. ], batch size: 100, lr: 9.78e-03, grad_scale: 8.0 2022-12-07 18:51:32,494 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.430e+02 3.290e+02 3.970e+02 9.558e+02, threshold=6.580e+02, percent-clipped=6.0 2022-12-07 18:51:38,698 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0582, 2.6348, 5.3023, 3.5122, 4.8078, 2.5061, 3.6706, 4.7955], device='cuda:1'), covar=tensor([0.0378, 0.4290, 0.0188, 0.7065, 0.0413, 0.3208, 0.1079, 0.0230], device='cuda:1'), in_proj_covar=tensor([0.0230, 0.0227, 0.0179, 0.0306, 0.0199, 0.0230, 0.0220, 0.0190], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:51:50,359 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.6722, 0.5785, 0.5936, 0.7332, 0.5296, 0.2949, 0.4309, 0.4717], device='cuda:1'), covar=tensor([0.0107, 0.0099, 0.0081, 0.0090, 0.0129, 0.0320, 0.0185, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0018, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.5024e-05, 9.1610e-05, 8.1767e-05, 8.7236e-05, 8.4088e-05, 1.2468e-04, 1.0460e-04, 1.1926e-04], device='cuda:1') 2022-12-07 18:52:05,589 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0149, 2.1284, 1.9699, 2.2012, 1.7446, 2.0258, 2.0921, 2.1170], device='cuda:1'), covar=tensor([0.1072, 0.1045, 0.1197, 0.0982, 0.1422, 0.0846, 0.1238, 0.0998], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0119, 0.0128, 0.0135, 0.0131, 0.0103, 0.0147, 0.0126], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 18:52:44,019 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-07 18:52:55,482 INFO [train.py:873] (1/4) Epoch 8, batch 5400, loss[loss=0.1624, simple_loss=0.1654, pruned_loss=0.07966, over 5975.00 frames. ], tot_loss[loss=0.1472, simple_loss=0.1708, pruned_loss=0.0618, over 1942362.67 frames. ], batch size: 100, lr: 9.77e-03, grad_scale: 8.0 2022-12-07 18:53:00,690 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.324e+02 2.944e+02 3.627e+02 6.890e+02, threshold=5.887e+02, percent-clipped=1.0 2022-12-07 18:53:36,470 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1168, 3.6493, 2.6734, 4.3397, 4.0558, 4.2625, 3.4146, 2.9966], device='cuda:1'), covar=tensor([0.0721, 0.1458, 0.4289, 0.0455, 0.0973, 0.0845, 0.1364, 0.3647], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0297, 0.0275, 0.0220, 0.0283, 0.0277, 0.0254, 0.0266], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 18:53:43,266 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:54:23,916 INFO [train.py:873] (1/4) Epoch 8, batch 5500, loss[loss=0.1351, simple_loss=0.1655, pruned_loss=0.05235, over 13938.00 frames. ], tot_loss[loss=0.1472, simple_loss=0.1708, pruned_loss=0.06182, over 2018933.36 frames. ], batch size: 20, lr: 9.76e-03, grad_scale: 8.0 2022-12-07 18:54:26,061 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:54:29,550 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.638e+01 2.350e+02 2.867e+02 3.886e+02 1.534e+03, threshold=5.734e+02, percent-clipped=10.0 2022-12-07 18:54:54,403 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8642, 1.6143, 2.1067, 1.6677, 1.9550, 1.4122, 1.6423, 1.8377], device='cuda:1'), covar=tensor([0.1595, 0.2247, 0.0273, 0.1210, 0.1010, 0.1620, 0.0886, 0.0517], device='cuda:1'), in_proj_covar=tensor([0.0232, 0.0232, 0.0182, 0.0311, 0.0203, 0.0236, 0.0223, 0.0191], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 18:55:18,753 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:55:21,720 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:55:52,341 INFO [train.py:873] (1/4) Epoch 8, batch 5600, loss[loss=0.125, simple_loss=0.1546, pruned_loss=0.04768, over 13937.00 frames. ], tot_loss[loss=0.1479, simple_loss=0.1708, pruned_loss=0.06247, over 1956679.03 frames. ], batch size: 19, lr: 9.75e-03, grad_scale: 8.0 2022-12-07 18:55:57,836 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.292e+02 2.814e+02 3.387e+02 5.498e+02, threshold=5.627e+02, percent-clipped=0.0 2022-12-07 18:56:03,690 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 18:57:21,321 INFO [train.py:873] (1/4) Epoch 8, batch 5700, loss[loss=0.1774, simple_loss=0.1905, pruned_loss=0.0821, over 11977.00 frames. ], tot_loss[loss=0.1487, simple_loss=0.171, pruned_loss=0.06323, over 1945569.97 frames. ], batch size: 100, lr: 9.75e-03, grad_scale: 8.0 2022-12-07 18:57:26,806 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.316e+02 2.828e+02 3.502e+02 7.696e+02, threshold=5.656e+02, percent-clipped=3.0 2022-12-07 18:58:02,177 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:58:49,961 INFO [train.py:873] (1/4) Epoch 8, batch 5800, loss[loss=0.1269, simple_loss=0.1623, pruned_loss=0.04582, over 14289.00 frames. ], tot_loss[loss=0.1481, simple_loss=0.1711, pruned_loss=0.06254, over 1958448.26 frames. ], batch size: 25, lr: 9.74e-03, grad_scale: 8.0 2022-12-07 18:58:55,173 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.215e+02 2.766e+02 3.426e+02 6.991e+02, threshold=5.532e+02, percent-clipped=1.0 2022-12-07 18:58:56,240 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:25,787 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:28,230 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 18:59:45,969 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:48,851 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:12,106 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 19:00:19,697 INFO [train.py:873] (1/4) Epoch 8, batch 5900, loss[loss=0.1605, simple_loss=0.1741, pruned_loss=0.07344, over 11968.00 frames. ], tot_loss[loss=0.148, simple_loss=0.1708, pruned_loss=0.06262, over 1990092.66 frames. ], batch size: 100, lr: 9.73e-03, grad_scale: 8.0 2022-12-07 19:00:20,740 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:25,227 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.290e+02 2.920e+02 3.537e+02 8.146e+02, threshold=5.840e+02, percent-clipped=6.0 2022-12-07 19:00:29,005 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:31,568 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:32,786 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.03 vs. limit=5.0 2022-12-07 19:01:22,569 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:01:46,303 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 19:01:47,422 INFO [train.py:873] (1/4) Epoch 8, batch 6000, loss[loss=0.126, simple_loss=0.1541, pruned_loss=0.04901, over 13944.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.1714, pruned_loss=0.06273, over 1998596.66 frames. ], batch size: 20, lr: 9.72e-03, grad_scale: 8.0 2022-12-07 19:01:47,422 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 19:02:08,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3694, 2.2594, 5.0543, 4.6253, 4.4546, 5.2252, 4.8137, 5.1835], device='cuda:1'), covar=tensor([0.1285, 0.1232, 0.0070, 0.0122, 0.0164, 0.0080, 0.0078, 0.0076], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0158, 0.0118, 0.0162, 0.0138, 0.0132, 0.0111, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:02:10,414 INFO [train.py:905] (1/4) Epoch 8, validation: loss=0.1228, simple_loss=0.1649, pruned_loss=0.04039, over 857387.00 frames. 2022-12-07 19:02:10,414 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 19:02:16,061 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.300e+02 3.027e+02 3.751e+02 1.020e+03, threshold=6.055e+02, percent-clipped=3.0 2022-12-07 19:02:38,957 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:02:56,771 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-07 19:03:38,336 INFO [train.py:873] (1/4) Epoch 8, batch 6100, loss[loss=0.167, simple_loss=0.165, pruned_loss=0.08451, over 3867.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.171, pruned_loss=0.06289, over 1893041.89 frames. ], batch size: 100, lr: 9.71e-03, grad_scale: 8.0 2022-12-07 19:03:40,173 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:03:43,429 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.326e+02 2.830e+02 3.436e+02 6.113e+02, threshold=5.661e+02, percent-clipped=1.0 2022-12-07 19:03:49,677 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0653, 1.8666, 0.9253, 1.4401, 1.3970, 1.1841, 1.2204, 1.2977], device='cuda:1'), covar=tensor([0.0593, 0.0523, 0.0824, 0.0701, 0.0752, 0.0548, 0.0585, 0.0873], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.6978e-05, 9.1755e-05, 8.2012e-05, 8.7856e-05, 8.4641e-05, 1.2731e-04, 1.0703e-04, 1.2101e-04], device='cuda:1') 2022-12-07 19:04:42,934 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6871, 2.0826, 3.7384, 2.5908, 3.6216, 1.9864, 2.7910, 3.4839], device='cuda:1'), covar=tensor([0.0551, 0.4119, 0.0400, 0.6789, 0.0459, 0.3420, 0.1259, 0.0424], device='cuda:1'), in_proj_covar=tensor([0.0234, 0.0231, 0.0184, 0.0316, 0.0204, 0.0234, 0.0224, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:04:42,979 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3592, 2.1522, 3.2921, 3.3956, 3.3219, 2.2924, 3.1443, 2.4644], device='cuda:1'), covar=tensor([0.0208, 0.0519, 0.0397, 0.0279, 0.0216, 0.0743, 0.0175, 0.0583], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0229, 0.0345, 0.0288, 0.0231, 0.0275, 0.0250, 0.0262], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:04:43,530 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 19:05:00,154 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2592, 1.9906, 2.2111, 1.5484, 1.9177, 2.2666, 2.3281, 1.9545], device='cuda:1'), covar=tensor([0.0807, 0.0797, 0.0955, 0.1698, 0.1101, 0.0643, 0.0481, 0.1600], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0185, 0.0127, 0.0124, 0.0121, 0.0127, 0.0104, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 19:05:02,946 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:05:06,212 INFO [train.py:873] (1/4) Epoch 8, batch 6200, loss[loss=0.1433, simple_loss=0.1364, pruned_loss=0.07503, over 2582.00 frames. ], tot_loss[loss=0.1479, simple_loss=0.1709, pruned_loss=0.06242, over 1972995.82 frames. ], batch size: 100, lr: 9.71e-03, grad_scale: 8.0 2022-12-07 19:05:11,776 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.400e+01 2.402e+02 3.111e+02 3.796e+02 6.098e+02, threshold=6.221e+02, percent-clipped=5.0 2022-12-07 19:05:32,133 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:06:26,929 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:06:35,505 INFO [train.py:873] (1/4) Epoch 8, batch 6300, loss[loss=0.1476, simple_loss=0.1682, pruned_loss=0.06356, over 14250.00 frames. ], tot_loss[loss=0.1467, simple_loss=0.1706, pruned_loss=0.0614, over 2026921.21 frames. ], batch size: 80, lr: 9.70e-03, grad_scale: 8.0 2022-12-07 19:06:40,625 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.206e+02 2.756e+02 3.412e+02 6.293e+02, threshold=5.512e+02, percent-clipped=2.0 2022-12-07 19:06:44,389 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3958, 1.7195, 1.2092, 1.4700, 1.2707, 1.0767, 1.1109, 1.0196], device='cuda:1'), covar=tensor([0.0554, 0.0504, 0.0446, 0.0328, 0.0609, 0.0296, 0.0365, 0.0825], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0020], device='cuda:1'), out_proj_covar=tensor([8.9981e-05, 9.6724e-05, 8.5014e-05, 9.0823e-05, 8.7574e-05, 1.3262e-04, 1.1108e-04, 1.2567e-04], device='cuda:1') 2022-12-07 19:07:00,036 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:07:26,870 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-07 19:08:04,133 INFO [train.py:873] (1/4) Epoch 8, batch 6400, loss[loss=0.1708, simple_loss=0.1824, pruned_loss=0.07955, over 10346.00 frames. ], tot_loss[loss=0.1451, simple_loss=0.1693, pruned_loss=0.06038, over 2019157.07 frames. ], batch size: 100, lr: 9.69e-03, grad_scale: 8.0 2022-12-07 19:08:06,075 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:08:09,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.267e+01 2.304e+02 2.776e+02 3.758e+02 5.907e+02, threshold=5.553e+02, percent-clipped=4.0 2022-12-07 19:08:24,824 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4760, 1.0045, 1.3253, 0.8920, 1.2162, 1.4442, 1.2019, 1.1632], device='cuda:1'), covar=tensor([0.0312, 0.0899, 0.0476, 0.0534, 0.0869, 0.0449, 0.0367, 0.0964], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0190, 0.0130, 0.0127, 0.0124, 0.0129, 0.0107, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:1') 2022-12-07 19:08:48,454 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:08:52,967 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5934, 2.4595, 2.1181, 2.2282, 2.5024, 2.4968, 2.5334, 2.5271], device='cuda:1'), covar=tensor([0.1153, 0.0984, 0.2451, 0.3060, 0.1017, 0.1205, 0.1465, 0.1086], device='cuda:1'), in_proj_covar=tensor([0.0336, 0.0239, 0.0397, 0.0509, 0.0285, 0.0375, 0.0368, 0.0323], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:09:29,264 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:09:32,600 INFO [train.py:873] (1/4) Epoch 8, batch 6500, loss[loss=0.1426, simple_loss=0.1689, pruned_loss=0.05812, over 14301.00 frames. ], tot_loss[loss=0.1468, simple_loss=0.1701, pruned_loss=0.06173, over 1963169.88 frames. ], batch size: 66, lr: 9.68e-03, grad_scale: 8.0 2022-12-07 19:09:37,914 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.343e+02 2.966e+02 3.672e+02 1.051e+03, threshold=5.932e+02, percent-clipped=2.0 2022-12-07 19:09:48,739 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.82 vs. limit=5.0 2022-12-07 19:10:10,832 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:17,823 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8526, 1.0610, 0.8005, 0.9174, 0.7142, 0.5549, 0.6242, 0.7389], device='cuda:1'), covar=tensor([0.0188, 0.0188, 0.0172, 0.0139, 0.0255, 0.0458, 0.0337, 0.0469], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.8279e-05, 9.5935e-05, 8.4581e-05, 9.0112e-05, 8.6912e-05, 1.3143e-04, 1.0984e-04, 1.2526e-04], device='cuda:1') 2022-12-07 19:10:21,460 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 19:10:31,173 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:40,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6076, 2.3896, 2.5042, 1.6499, 2.2149, 2.4389, 2.7212, 2.2879], device='cuda:1'), covar=tensor([0.1024, 0.1539, 0.1193, 0.2221, 0.1140, 0.0735, 0.0717, 0.1619], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0188, 0.0129, 0.0126, 0.0122, 0.0129, 0.0106, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:1') 2022-12-07 19:10:46,630 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:59,606 INFO [train.py:873] (1/4) Epoch 8, batch 6600, loss[loss=0.1425, simple_loss=0.1702, pruned_loss=0.05741, over 14009.00 frames. ], tot_loss[loss=0.1476, simple_loss=0.1704, pruned_loss=0.06241, over 1946967.79 frames. ], batch size: 19, lr: 9.67e-03, grad_scale: 8.0 2022-12-07 19:11:04,956 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.283e+02 2.818e+02 3.412e+02 6.206e+02, threshold=5.635e+02, percent-clipped=2.0 2022-12-07 19:11:23,669 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:11:24,620 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:11:39,229 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 19:11:45,738 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 19:11:47,443 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 19:12:05,791 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:12:27,867 INFO [train.py:873] (1/4) Epoch 8, batch 6700, loss[loss=0.1635, simple_loss=0.1938, pruned_loss=0.06655, over 13986.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.1711, pruned_loss=0.06286, over 1950098.58 frames. ], batch size: 22, lr: 9.66e-03, grad_scale: 8.0 2022-12-07 19:12:32,694 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.361e+02 2.930e+02 3.597e+02 6.848e+02, threshold=5.861e+02, percent-clipped=3.0 2022-12-07 19:12:33,723 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9687, 4.5868, 4.2421, 4.5177, 4.4555, 4.8245, 4.8918, 4.9160], device='cuda:1'), covar=tensor([0.0704, 0.0521, 0.2311, 0.2616, 0.0792, 0.0690, 0.0907, 0.0779], device='cuda:1'), in_proj_covar=tensor([0.0340, 0.0242, 0.0401, 0.0511, 0.0289, 0.0378, 0.0367, 0.0328], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:13:11,602 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.14 vs. limit=2.0 2022-12-07 19:13:14,160 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=11.83 vs. limit=5.0 2022-12-07 19:13:17,595 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 19:13:36,370 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2579, 2.1983, 3.0960, 2.4548, 3.1381, 2.9790, 2.9514, 2.4319], device='cuda:1'), covar=tensor([0.0816, 0.3097, 0.1156, 0.2241, 0.0837, 0.0978, 0.1463, 0.2229], device='cuda:1'), in_proj_covar=tensor([0.0328, 0.0323, 0.0395, 0.0307, 0.0376, 0.0309, 0.0359, 0.0327], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:13:55,715 INFO [train.py:873] (1/4) Epoch 8, batch 6800, loss[loss=0.165, simple_loss=0.1799, pruned_loss=0.07498, over 9453.00 frames. ], tot_loss[loss=0.1485, simple_loss=0.1712, pruned_loss=0.06286, over 1901551.30 frames. ], batch size: 100, lr: 9.66e-03, grad_scale: 8.0 2022-12-07 19:14:01,555 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.278e+02 2.761e+02 3.768e+02 7.478e+02, threshold=5.523e+02, percent-clipped=3.0 2022-12-07 19:15:12,055 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:25,704 INFO [train.py:873] (1/4) Epoch 8, batch 6900, loss[loss=0.1406, simple_loss=0.1676, pruned_loss=0.05681, over 14019.00 frames. ], tot_loss[loss=0.1472, simple_loss=0.1701, pruned_loss=0.06212, over 1912548.63 frames. ], batch size: 20, lr: 9.65e-03, grad_scale: 8.0 2022-12-07 19:15:30,915 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.523e+02 3.039e+02 3.738e+02 1.408e+03, threshold=6.079e+02, percent-clipped=6.0 2022-12-07 19:15:36,987 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:45,686 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:54,677 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:13,173 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:31,391 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:53,932 INFO [train.py:873] (1/4) Epoch 8, batch 7000, loss[loss=0.09702, simple_loss=0.1329, pruned_loss=0.03059, over 13584.00 frames. ], tot_loss[loss=0.1473, simple_loss=0.1704, pruned_loss=0.06207, over 1931701.15 frames. ], batch size: 17, lr: 9.64e-03, grad_scale: 8.0 2022-12-07 19:16:59,264 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.338e+02 3.066e+02 3.722e+02 8.266e+02, threshold=6.131e+02, percent-clipped=2.0 2022-12-07 19:17:07,984 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:17:43,715 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2074, 3.9708, 3.7861, 4.1819, 4.0411, 3.6412, 4.2673, 3.5683], device='cuda:1'), covar=tensor([0.0423, 0.0812, 0.0373, 0.0453, 0.0684, 0.1226, 0.0514, 0.0515], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0247, 0.0170, 0.0163, 0.0164, 0.0133, 0.0251, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 19:18:27,445 INFO [train.py:873] (1/4) Epoch 8, batch 7100, loss[loss=0.1613, simple_loss=0.178, pruned_loss=0.07228, over 14239.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1696, pruned_loss=0.06135, over 1917486.73 frames. ], batch size: 69, lr: 9.63e-03, grad_scale: 8.0 2022-12-07 19:18:31,848 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:18:32,560 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.178e+02 2.684e+02 3.271e+02 7.336e+02, threshold=5.369e+02, percent-clipped=1.0 2022-12-07 19:18:42,268 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:18:54,092 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:05,211 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:24,581 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 19:19:35,590 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 19:19:46,374 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:53,836 INFO [train.py:873] (1/4) Epoch 8, batch 7200, loss[loss=0.1252, simple_loss=0.1536, pruned_loss=0.04839, over 14050.00 frames. ], tot_loss[loss=0.1476, simple_loss=0.1708, pruned_loss=0.06219, over 1986222.31 frames. ], batch size: 19, lr: 9.63e-03, grad_scale: 8.0 2022-12-07 19:19:57,775 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:59,258 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.579e+01 2.475e+02 2.936e+02 3.540e+02 1.198e+03, threshold=5.873e+02, percent-clipped=8.0 2022-12-07 19:20:14,443 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:24,584 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3828, 2.0276, 3.5200, 2.5475, 3.3635, 1.8690, 2.7204, 3.2357], device='cuda:1'), covar=tensor([0.0546, 0.4043, 0.0358, 0.5315, 0.0541, 0.3343, 0.1101, 0.0405], device='cuda:1'), in_proj_covar=tensor([0.0233, 0.0231, 0.0184, 0.0313, 0.0207, 0.0233, 0.0223, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:20:30,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0404, 1.6537, 2.0316, 1.4259, 1.6598, 2.1299, 1.9556, 1.7745], device='cuda:1'), covar=tensor([0.0768, 0.0830, 0.0782, 0.1456, 0.1323, 0.0709, 0.0797, 0.1472], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0184, 0.0128, 0.0124, 0.0123, 0.0129, 0.0104, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:1') 2022-12-07 19:20:49,271 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:49,675 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 19:20:54,641 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:56,408 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:21:22,474 INFO [train.py:873] (1/4) Epoch 8, batch 7300, loss[loss=0.1594, simple_loss=0.1603, pruned_loss=0.07926, over 3918.00 frames. ], tot_loss[loss=0.1473, simple_loss=0.1703, pruned_loss=0.06216, over 1994459.92 frames. ], batch size: 100, lr: 9.62e-03, grad_scale: 16.0 2022-12-07 19:21:28,208 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.538e+01 2.297e+02 3.128e+02 4.022e+02 9.031e+02, threshold=6.256e+02, percent-clipped=2.0 2022-12-07 19:21:30,682 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:21:43,216 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:22:32,328 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 19:22:40,217 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9341, 1.9329, 2.1624, 1.5073, 1.4660, 2.0623, 1.3300, 1.8932], device='cuda:1'), covar=tensor([0.0921, 0.1906, 0.0642, 0.1848, 0.2848, 0.0721, 0.3752, 0.0840], device='cuda:1'), in_proj_covar=tensor([0.0074, 0.0092, 0.0082, 0.0090, 0.0111, 0.0078, 0.0130, 0.0079], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:22:47,628 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 19:22:49,539 INFO [train.py:873] (1/4) Epoch 8, batch 7400, loss[loss=0.1597, simple_loss=0.1619, pruned_loss=0.07874, over 3830.00 frames. ], tot_loss[loss=0.147, simple_loss=0.1702, pruned_loss=0.06184, over 1992253.22 frames. ], batch size: 100, lr: 9.61e-03, grad_scale: 8.0 2022-12-07 19:22:55,005 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 19:22:56,039 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.246e+02 2.782e+02 3.435e+02 7.808e+02, threshold=5.565e+02, percent-clipped=3.0 2022-12-07 19:23:09,170 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 19:23:41,052 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2022-12-07 19:23:43,977 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 19:23:54,299 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 19:24:06,062 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:24:17,783 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:24:18,543 INFO [train.py:873] (1/4) Epoch 8, batch 7500, loss[loss=0.1516, simple_loss=0.179, pruned_loss=0.06213, over 14294.00 frames. ], tot_loss[loss=0.1478, simple_loss=0.1709, pruned_loss=0.06234, over 2009433.23 frames. ], batch size: 76, lr: 9.60e-03, grad_scale: 8.0 2022-12-07 19:24:24,272 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.510e+02 2.850e+02 3.552e+02 6.747e+02, threshold=5.701e+02, percent-clipped=5.0 2022-12-07 19:24:32,102 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8766, 1.5277, 3.9853, 3.6569, 3.7712, 4.0938, 3.3954, 4.0776], device='cuda:1'), covar=tensor([0.1289, 0.1434, 0.0085, 0.0197, 0.0173, 0.0090, 0.0183, 0.0094], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0158, 0.0119, 0.0164, 0.0139, 0.0132, 0.0112, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:25:46,335 INFO [train.py:873] (1/4) Epoch 9, batch 0, loss[loss=0.1664, simple_loss=0.1881, pruned_loss=0.07235, over 14231.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1881, pruned_loss=0.07235, over 14231.00 frames. ], batch size: 60, lr: 9.08e-03, grad_scale: 8.0 2022-12-07 19:25:46,335 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 19:25:53,548 INFO [train.py:905] (1/4) Epoch 9, validation: loss=0.1275, simple_loss=0.1706, pruned_loss=0.04216, over 857387.00 frames. 2022-12-07 19:25:53,549 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 19:26:00,174 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:33,659 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.503e+01 2.217e+02 2.837e+02 3.988e+02 6.329e+02, threshold=5.675e+02, percent-clipped=3.0 2022-12-07 19:26:36,532 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:43,323 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:45,466 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:27:19,870 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:27:23,501 INFO [train.py:873] (1/4) Epoch 9, batch 100, loss[loss=0.1304, simple_loss=0.1579, pruned_loss=0.05148, over 14247.00 frames. ], tot_loss[loss=0.145, simple_loss=0.1699, pruned_loss=0.06009, over 943963.07 frames. ], batch size: 32, lr: 9.08e-03, grad_scale: 8.0 2022-12-07 19:27:58,017 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2609, 2.9434, 2.9668, 3.1914, 3.1449, 3.2314, 3.2811, 2.7063], device='cuda:1'), covar=tensor([0.0571, 0.1167, 0.0493, 0.0577, 0.0786, 0.0421, 0.0669, 0.0670], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0243, 0.0165, 0.0161, 0.0162, 0.0132, 0.0249, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 19:28:03,331 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.443e+02 2.952e+02 3.884e+02 8.156e+02, threshold=5.905e+02, percent-clipped=9.0 2022-12-07 19:28:04,998 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 19:28:51,074 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:28:51,887 INFO [train.py:873] (1/4) Epoch 9, batch 200, loss[loss=0.1192, simple_loss=0.1486, pruned_loss=0.04493, over 14184.00 frames. ], tot_loss[loss=0.1458, simple_loss=0.1696, pruned_loss=0.06096, over 1316064.24 frames. ], batch size: 37, lr: 9.07e-03, grad_scale: 8.0 2022-12-07 19:29:01,480 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:04,028 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=10.21 vs. limit=5.0 2022-12-07 19:29:10,918 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:13,754 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:24,903 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:32,172 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.211e+01 2.496e+02 3.042e+02 4.274e+02 1.188e+03, threshold=6.084e+02, percent-clipped=4.0 2022-12-07 19:29:33,117 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:44,424 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:55,853 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:04,752 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:07,105 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:20,885 INFO [train.py:873] (1/4) Epoch 9, batch 300, loss[loss=0.1181, simple_loss=0.147, pruned_loss=0.04462, over 14274.00 frames. ], tot_loss[loss=0.145, simple_loss=0.1692, pruned_loss=0.06042, over 1600361.13 frames. ], batch size: 25, lr: 9.06e-03, grad_scale: 4.0 2022-12-07 19:31:01,440 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.299e+02 2.848e+02 3.507e+02 1.292e+03, threshold=5.696e+02, percent-clipped=2.0 2022-12-07 19:31:11,057 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:31:49,308 INFO [train.py:873] (1/4) Epoch 9, batch 400, loss[loss=0.1423, simple_loss=0.1386, pruned_loss=0.07305, over 2562.00 frames. ], tot_loss[loss=0.1442, simple_loss=0.1684, pruned_loss=0.05999, over 1722267.04 frames. ], batch size: 100, lr: 9.06e-03, grad_scale: 8.0 2022-12-07 19:31:53,805 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:31:57,284 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1445, 1.2638, 1.4177, 0.8986, 0.8924, 1.2843, 0.8364, 1.1848], device='cuda:1'), covar=tensor([0.1589, 0.2381, 0.0806, 0.2162, 0.2940, 0.0897, 0.2431, 0.1084], device='cuda:1'), in_proj_covar=tensor([0.0074, 0.0091, 0.0084, 0.0088, 0.0109, 0.0077, 0.0128, 0.0079], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:32:09,128 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8520, 1.2820, 2.0904, 1.2568, 2.0026, 2.0553, 1.8178, 2.0778], device='cuda:1'), covar=tensor([0.0344, 0.1707, 0.0356, 0.1632, 0.0410, 0.0457, 0.0687, 0.0332], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0156, 0.0151, 0.0165, 0.0166, 0.0162, 0.0132, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:32:13,042 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 19:32:29,748 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.639e+01 2.151e+02 2.975e+02 3.759e+02 9.050e+02, threshold=5.950e+02, percent-clipped=6.0 2022-12-07 19:32:35,915 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8972, 1.3333, 3.0856, 2.8387, 2.9946, 3.1209, 2.3853, 3.0748], device='cuda:1'), covar=tensor([0.1112, 0.1325, 0.0131, 0.0287, 0.0268, 0.0133, 0.0361, 0.0174], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0158, 0.0120, 0.0164, 0.0140, 0.0134, 0.0113, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:33:01,326 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2971, 1.4849, 1.3056, 1.1310, 1.1648, 0.8399, 0.8694, 0.9133], device='cuda:1'), covar=tensor([0.0215, 0.0206, 0.0222, 0.0247, 0.0308, 0.0341, 0.0212, 0.0421], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:1'), out_proj_covar=tensor([8.9671e-05, 9.5886e-05, 8.4972e-05, 9.0581e-05, 8.6427e-05, 1.3197e-04, 1.1040e-04, 1.2410e-04], device='cuda:1') 2022-12-07 19:33:18,101 INFO [train.py:873] (1/4) Epoch 9, batch 500, loss[loss=0.1163, simple_loss=0.1534, pruned_loss=0.0396, over 13896.00 frames. ], tot_loss[loss=0.1451, simple_loss=0.1689, pruned_loss=0.06061, over 1842575.02 frames. ], batch size: 20, lr: 9.05e-03, grad_scale: 8.0 2022-12-07 19:33:59,480 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 2.291e+02 2.820e+02 3.548e+02 5.650e+02, threshold=5.639e+02, percent-clipped=0.0 2022-12-07 19:34:14,688 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:34:25,970 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:34:27,182 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9789, 0.9647, 0.9689, 0.9919, 1.1377, 0.5187, 0.8880, 1.0013], device='cuda:1'), covar=tensor([0.0537, 0.1083, 0.0650, 0.0608, 0.0454, 0.0928, 0.0762, 0.0697], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0024, 0.0024, 0.0022, 0.0023, 0.0033, 0.0023, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 19:34:38,048 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 19:34:39,832 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2022-12-07 19:34:45,879 INFO [train.py:873] (1/4) Epoch 9, batch 600, loss[loss=0.1799, simple_loss=0.1907, pruned_loss=0.08456, over 4987.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1697, pruned_loss=0.06137, over 1925475.95 frames. ], batch size: 100, lr: 9.04e-03, grad_scale: 8.0 2022-12-07 19:35:08,460 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:35:25,943 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.178e+01 2.284e+02 2.726e+02 3.406e+02 8.336e+02, threshold=5.452e+02, percent-clipped=5.0 2022-12-07 19:35:48,849 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 19:35:59,628 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2022-12-07 19:36:13,113 INFO [train.py:873] (1/4) Epoch 9, batch 700, loss[loss=0.1411, simple_loss=0.1663, pruned_loss=0.05795, over 14233.00 frames. ], tot_loss[loss=0.1443, simple_loss=0.1683, pruned_loss=0.06012, over 1979905.60 frames. ], batch size: 69, lr: 9.03e-03, grad_scale: 8.0 2022-12-07 19:36:27,875 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2554, 2.2777, 2.3961, 2.5218, 1.9989, 2.4854, 2.1575, 1.1534], device='cuda:1'), covar=tensor([0.1547, 0.0925, 0.0935, 0.0410, 0.1096, 0.0626, 0.1473, 0.2818], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0068, 0.0055, 0.0056, 0.0086, 0.0064, 0.0089, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 19:36:54,420 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.220e+02 2.923e+02 3.750e+02 6.888e+02, threshold=5.845e+02, percent-clipped=6.0 2022-12-07 19:36:58,399 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 19:37:01,064 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.41 vs. limit=5.0 2022-12-07 19:37:10,133 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-07 19:37:41,703 INFO [train.py:873] (1/4) Epoch 9, batch 800, loss[loss=0.1921, simple_loss=0.1918, pruned_loss=0.09626, over 7799.00 frames. ], tot_loss[loss=0.1451, simple_loss=0.169, pruned_loss=0.06056, over 1968660.42 frames. ], batch size: 100, lr: 9.03e-03, grad_scale: 8.0 2022-12-07 19:38:02,408 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7304, 2.0072, 2.0008, 2.1268, 1.9244, 2.2033, 1.7408, 1.2397], device='cuda:1'), covar=tensor([0.1542, 0.0759, 0.0854, 0.0460, 0.1009, 0.0475, 0.1586, 0.2517], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0067, 0.0055, 0.0056, 0.0086, 0.0064, 0.0090, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 19:38:14,368 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8689, 2.9483, 3.0554, 2.9926, 2.9231, 2.8404, 1.3571, 2.7387], device='cuda:1'), covar=tensor([0.0395, 0.0406, 0.0464, 0.0426, 0.0398, 0.0794, 0.3026, 0.0353], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0155, 0.0132, 0.0129, 0.0184, 0.0126, 0.0152, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:38:22,022 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 2.254e+02 2.833e+02 3.522e+02 6.123e+02, threshold=5.665e+02, percent-clipped=1.0 2022-12-07 19:38:49,231 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:38:53,622 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6135, 1.4969, 3.5680, 1.6884, 3.5616, 3.6638, 2.5179, 4.0142], device='cuda:1'), covar=tensor([0.0244, 0.2997, 0.0403, 0.2249, 0.0564, 0.0419, 0.0943, 0.0149], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0159, 0.0157, 0.0172, 0.0170, 0.0169, 0.0137, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:39:09,293 INFO [train.py:873] (1/4) Epoch 9, batch 900, loss[loss=0.1546, simple_loss=0.175, pruned_loss=0.06705, over 14182.00 frames. ], tot_loss[loss=0.1456, simple_loss=0.1692, pruned_loss=0.06097, over 1997574.32 frames. ], batch size: 89, lr: 9.02e-03, grad_scale: 8.0 2022-12-07 19:39:27,291 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:39:31,533 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:39:37,102 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-07 19:39:50,649 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.290e+02 3.289e+02 4.035e+02 9.123e+02, threshold=6.578e+02, percent-clipped=5.0 2022-12-07 19:40:24,390 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2349, 1.4340, 3.3923, 1.3973, 3.0826, 3.2959, 2.2106, 3.5592], device='cuda:1'), covar=tensor([0.0207, 0.2685, 0.0287, 0.2257, 0.0804, 0.0361, 0.0977, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0159, 0.0157, 0.0172, 0.0170, 0.0170, 0.0138, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:40:33,258 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3423, 2.4187, 4.4558, 3.0164, 4.2157, 2.0652, 3.4218, 4.1175], device='cuda:1'), covar=tensor([0.0539, 0.3935, 0.0300, 0.7157, 0.0523, 0.3692, 0.1150, 0.0356], device='cuda:1'), in_proj_covar=tensor([0.0239, 0.0227, 0.0186, 0.0314, 0.0212, 0.0230, 0.0220, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:40:35,648 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7262, 3.3976, 3.3483, 3.7019, 3.5549, 3.6465, 3.6736, 3.0930], device='cuda:1'), covar=tensor([0.0359, 0.0987, 0.0384, 0.0425, 0.0632, 0.0338, 0.0597, 0.0542], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0241, 0.0166, 0.0162, 0.0161, 0.0131, 0.0249, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 19:40:37,279 INFO [train.py:873] (1/4) Epoch 9, batch 1000, loss[loss=0.1222, simple_loss=0.1549, pruned_loss=0.0448, over 14539.00 frames. ], tot_loss[loss=0.1474, simple_loss=0.1701, pruned_loss=0.06234, over 1965304.74 frames. ], batch size: 24, lr: 9.01e-03, grad_scale: 8.0 2022-12-07 19:41:09,266 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 19:41:18,075 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.101e+02 2.828e+02 3.770e+02 7.982e+02, threshold=5.656e+02, percent-clipped=1.0 2022-12-07 19:42:05,669 INFO [train.py:873] (1/4) Epoch 9, batch 1100, loss[loss=0.1428, simple_loss=0.1713, pruned_loss=0.0571, over 14307.00 frames. ], tot_loss[loss=0.1468, simple_loss=0.17, pruned_loss=0.06178, over 1965486.46 frames. ], batch size: 46, lr: 9.00e-03, grad_scale: 8.0 2022-12-07 19:42:46,207 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.543e+02 3.100e+02 3.894e+02 7.246e+02, threshold=6.200e+02, percent-clipped=7.0 2022-12-07 19:43:32,679 INFO [train.py:873] (1/4) Epoch 9, batch 1200, loss[loss=0.1934, simple_loss=0.1828, pruned_loss=0.1019, over 3862.00 frames. ], tot_loss[loss=0.1456, simple_loss=0.1693, pruned_loss=0.06096, over 1950674.10 frames. ], batch size: 100, lr: 9.00e-03, grad_scale: 8.0 2022-12-07 19:43:49,900 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:43:50,300 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 19:44:03,906 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:13,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.189e+02 2.580e+02 3.174e+02 5.497e+02, threshold=5.160e+02, percent-clipped=0.0 2022-12-07 19:44:13,461 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:32,973 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:56,433 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:59,741 INFO [train.py:873] (1/4) Epoch 9, batch 1300, loss[loss=0.1735, simple_loss=0.1546, pruned_loss=0.09619, over 1177.00 frames. ], tot_loss[loss=0.1458, simple_loss=0.1693, pruned_loss=0.06109, over 1894548.85 frames. ], batch size: 100, lr: 8.99e-03, grad_scale: 8.0 2022-12-07 19:45:07,086 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:45:17,638 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8778, 1.6175, 3.7178, 1.5452, 3.7244, 3.8650, 2.7496, 4.2261], device='cuda:1'), covar=tensor([0.0188, 0.3024, 0.0441, 0.2477, 0.0464, 0.0330, 0.0744, 0.0129], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0159, 0.0156, 0.0172, 0.0170, 0.0170, 0.0135, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:45:26,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 19:45:41,028 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.302e+02 2.881e+02 3.480e+02 7.451e+02, threshold=5.762e+02, percent-clipped=5.0 2022-12-07 19:45:56,531 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 19:46:28,404 INFO [train.py:873] (1/4) Epoch 9, batch 1400, loss[loss=0.1334, simple_loss=0.1662, pruned_loss=0.05033, over 14260.00 frames. ], tot_loss[loss=0.146, simple_loss=0.1696, pruned_loss=0.06114, over 1936005.51 frames. ], batch size: 28, lr: 8.98e-03, grad_scale: 8.0 2022-12-07 19:46:42,949 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:47:08,293 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.184e+02 2.770e+02 3.491e+02 7.127e+02, threshold=5.539e+02, percent-clipped=2.0 2022-12-07 19:47:37,227 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:47:40,639 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5670, 1.2927, 3.2672, 3.1483, 3.2155, 3.3356, 2.5925, 3.2444], device='cuda:1'), covar=tensor([0.2128, 0.2279, 0.0278, 0.0430, 0.0414, 0.0322, 0.0699, 0.0329], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0156, 0.0120, 0.0160, 0.0139, 0.0132, 0.0115, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:47:51,795 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8732, 1.9196, 2.1285, 1.3922, 1.4643, 1.9463, 1.1496, 1.8347], device='cuda:1'), covar=tensor([0.1572, 0.1898, 0.0828, 0.2774, 0.3299, 0.1026, 0.4915, 0.1226], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0090, 0.0083, 0.0089, 0.0109, 0.0076, 0.0127, 0.0080], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:47:56,373 INFO [train.py:873] (1/4) Epoch 9, batch 1500, loss[loss=0.182, simple_loss=0.1632, pruned_loss=0.1004, over 1223.00 frames. ], tot_loss[loss=0.1442, simple_loss=0.1683, pruned_loss=0.06009, over 1943859.09 frames. ], batch size: 100, lr: 8.98e-03, grad_scale: 8.0 2022-12-07 19:48:37,573 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.345e+02 2.854e+02 3.466e+02 6.884e+02, threshold=5.708e+02, percent-clipped=1.0 2022-12-07 19:48:42,384 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1807, 2.1723, 3.1213, 3.2112, 3.0869, 2.2429, 3.1391, 2.5564], device='cuda:1'), covar=tensor([0.0253, 0.0583, 0.0490, 0.0265, 0.0273, 0.0827, 0.0243, 0.0600], device='cuda:1'), in_proj_covar=tensor([0.0257, 0.0235, 0.0348, 0.0293, 0.0237, 0.0282, 0.0261, 0.0264], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:49:17,141 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 19:49:18,221 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:49:18,272 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1416, 2.0672, 2.4587, 1.3904, 1.6669, 2.2159, 1.3955, 2.1294], device='cuda:1'), covar=tensor([0.1067, 0.1876, 0.0661, 0.2792, 0.2643, 0.0826, 0.4017, 0.0977], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0091, 0.0082, 0.0090, 0.0108, 0.0077, 0.0128, 0.0081], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:49:23,354 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9266, 2.1084, 2.7387, 2.3345, 2.7291, 2.6203, 2.7013, 2.4092], device='cuda:1'), covar=tensor([0.0623, 0.2870, 0.0789, 0.1901, 0.0681, 0.0831, 0.1299, 0.1508], device='cuda:1'), in_proj_covar=tensor([0.0331, 0.0322, 0.0395, 0.0312, 0.0375, 0.0310, 0.0364, 0.0328], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:49:25,698 INFO [train.py:873] (1/4) Epoch 9, batch 1600, loss[loss=0.1444, simple_loss=0.1752, pruned_loss=0.05682, over 14284.00 frames. ], tot_loss[loss=0.1447, simple_loss=0.1686, pruned_loss=0.06042, over 1994005.78 frames. ], batch size: 39, lr: 8.97e-03, grad_scale: 8.0 2022-12-07 19:49:27,554 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:49:48,692 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:49:55,733 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8913, 1.2506, 2.0946, 1.1943, 2.0265, 2.0307, 1.7187, 2.1277], device='cuda:1'), covar=tensor([0.0328, 0.1800, 0.0322, 0.1865, 0.0417, 0.0505, 0.0865, 0.0329], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0157, 0.0154, 0.0170, 0.0167, 0.0167, 0.0134, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 19:50:07,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.173e+02 2.069e+02 2.487e+02 3.323e+02 7.707e+02, threshold=4.973e+02, percent-clipped=4.0 2022-12-07 19:50:42,667 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:50:54,728 INFO [train.py:873] (1/4) Epoch 9, batch 1700, loss[loss=0.1517, simple_loss=0.1682, pruned_loss=0.06758, over 9487.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1687, pruned_loss=0.06045, over 1946562.53 frames. ], batch size: 100, lr: 8.96e-03, grad_scale: 8.0 2022-12-07 19:51:18,174 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7579, 2.3433, 3.5386, 2.5797, 3.5404, 3.4888, 3.3979, 2.8599], device='cuda:1'), covar=tensor([0.0680, 0.3684, 0.0861, 0.2275, 0.0671, 0.0939, 0.1615, 0.2865], device='cuda:1'), in_proj_covar=tensor([0.0328, 0.0318, 0.0393, 0.0309, 0.0369, 0.0307, 0.0358, 0.0323], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:51:35,608 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.316e+02 3.145e+02 3.790e+02 7.968e+02, threshold=6.289e+02, percent-clipped=7.0 2022-12-07 19:51:59,384 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:52:01,943 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:52:09,885 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 19:52:23,523 INFO [train.py:873] (1/4) Epoch 9, batch 1800, loss[loss=0.1951, simple_loss=0.1994, pruned_loss=0.09538, over 10366.00 frames. ], tot_loss[loss=0.1451, simple_loss=0.1691, pruned_loss=0.06056, over 1954021.62 frames. ], batch size: 100, lr: 8.95e-03, grad_scale: 8.0 2022-12-07 19:52:56,752 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:04,486 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.235e+02 2.778e+02 3.404e+02 6.725e+02, threshold=5.556e+02, percent-clipped=1.0 2022-12-07 19:53:16,178 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2022-12-07 19:53:34,130 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0052, 1.9230, 2.0420, 2.0925, 2.0143, 1.6384, 1.2796, 1.8180], device='cuda:1'), covar=tensor([0.0411, 0.0451, 0.0500, 0.0235, 0.0409, 0.1054, 0.1999, 0.0379], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0154, 0.0129, 0.0125, 0.0184, 0.0124, 0.0151, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:53:44,461 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:47,022 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:52,234 INFO [train.py:873] (1/4) Epoch 9, batch 1900, loss[loss=0.1237, simple_loss=0.1568, pruned_loss=0.04531, over 14366.00 frames. ], tot_loss[loss=0.1436, simple_loss=0.1684, pruned_loss=0.05944, over 1947190.63 frames. ], batch size: 73, lr: 8.95e-03, grad_scale: 8.0 2022-12-07 19:53:54,362 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:54:26,828 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:54:28,777 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6517, 1.5922, 1.5778, 1.7590, 1.3314, 1.2786, 1.4144, 1.5003], device='cuda:1'), covar=tensor([0.0393, 0.0569, 0.1037, 0.0368, 0.0676, 0.0504, 0.0547, 0.0876], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:1'), out_proj_covar=tensor([9.2038e-05, 9.9633e-05, 8.8265e-05, 9.6706e-05, 9.1003e-05, 1.3755e-04, 1.1700e-04, 1.3138e-04], device='cuda:1') 2022-12-07 19:54:33,571 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.190e+02 2.770e+02 3.608e+02 5.661e+02, threshold=5.540e+02, percent-clipped=1.0 2022-12-07 19:54:37,190 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:54:41,696 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:55:04,155 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:55:17,801 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3077, 2.2379, 3.1063, 2.4887, 3.1575, 3.0392, 3.0235, 2.5881], device='cuda:1'), covar=tensor([0.1060, 0.2857, 0.0864, 0.2009, 0.0955, 0.0724, 0.1487, 0.2123], device='cuda:1'), in_proj_covar=tensor([0.0328, 0.0318, 0.0391, 0.0308, 0.0369, 0.0305, 0.0356, 0.0322], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:55:20,939 INFO [train.py:873] (1/4) Epoch 9, batch 2000, loss[loss=0.1838, simple_loss=0.1787, pruned_loss=0.09447, over 3866.00 frames. ], tot_loss[loss=0.1438, simple_loss=0.1681, pruned_loss=0.05974, over 1902634.83 frames. ], batch size: 100, lr: 8.94e-03, grad_scale: 8.0 2022-12-07 19:55:29,915 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4241, 3.7262, 3.1398, 4.7064, 4.2484, 4.4732, 3.7877, 3.3261], device='cuda:1'), covar=tensor([0.0670, 0.1406, 0.3658, 0.0404, 0.0809, 0.1156, 0.1204, 0.3108], device='cuda:1'), in_proj_covar=tensor([0.0257, 0.0296, 0.0279, 0.0234, 0.0294, 0.0280, 0.0255, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 19:55:40,588 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 19:56:01,043 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 2.315e+02 2.841e+02 3.698e+02 9.209e+02, threshold=5.682e+02, percent-clipped=7.0 2022-12-07 19:56:03,021 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:10,507 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6912, 1.3671, 2.4358, 2.2013, 2.3279, 2.3679, 1.5771, 2.3995], device='cuda:1'), covar=tensor([0.1100, 0.1450, 0.0213, 0.0538, 0.0459, 0.0229, 0.0891, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0155, 0.0119, 0.0160, 0.0139, 0.0132, 0.0114, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 19:56:12,694 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 19:56:24,933 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:37,545 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:37,579 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4057, 1.9345, 3.5966, 2.5482, 3.4789, 1.9186, 2.7992, 3.3430], device='cuda:1'), covar=tensor([0.0628, 0.4460, 0.0375, 0.5883, 0.0617, 0.3533, 0.1125, 0.0488], device='cuda:1'), in_proj_covar=tensor([0.0238, 0.0230, 0.0189, 0.0309, 0.0211, 0.0231, 0.0220, 0.0190], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:56:47,847 INFO [train.py:873] (1/4) Epoch 9, batch 2100, loss[loss=0.1494, simple_loss=0.1682, pruned_loss=0.06528, over 5953.00 frames. ], tot_loss[loss=0.1428, simple_loss=0.1673, pruned_loss=0.05911, over 1897874.95 frames. ], batch size: 100, lr: 8.93e-03, grad_scale: 8.0 2022-12-07 19:56:56,326 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:00,069 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7704, 4.4405, 4.3351, 4.7647, 4.3609, 3.9183, 4.8090, 4.6130], device='cuda:1'), covar=tensor([0.0597, 0.0690, 0.0743, 0.0560, 0.0766, 0.0689, 0.0571, 0.0773], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0120, 0.0131, 0.0138, 0.0134, 0.0107, 0.0150, 0.0129], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 19:57:07,119 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:16,347 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:28,900 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.143e+02 2.774e+02 3.519e+02 5.598e+02, threshold=5.548e+02, percent-clipped=0.0 2022-12-07 19:57:31,558 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:48,830 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:58:15,319 INFO [train.py:873] (1/4) Epoch 9, batch 2200, loss[loss=0.1772, simple_loss=0.1595, pruned_loss=0.09744, over 1196.00 frames. ], tot_loss[loss=0.1443, simple_loss=0.1681, pruned_loss=0.06028, over 1918864.18 frames. ], batch size: 100, lr: 8.93e-03, grad_scale: 8.0 2022-12-07 19:58:42,067 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:58:45,745 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0060, 4.7909, 4.3694, 4.6294, 4.6687, 4.9027, 5.0503, 4.9724], device='cuda:1'), covar=tensor([0.0807, 0.0361, 0.2077, 0.2613, 0.0706, 0.0724, 0.0698, 0.0829], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0241, 0.0410, 0.0516, 0.0298, 0.0386, 0.0374, 0.0338], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 19:58:56,803 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.557e+02 3.266e+02 4.324e+02 1.294e+03, threshold=6.533e+02, percent-clipped=13.0 2022-12-07 19:58:59,485 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:21,889 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:25,949 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:41,694 INFO [train.py:873] (1/4) Epoch 9, batch 2300, loss[loss=0.1291, simple_loss=0.1575, pruned_loss=0.05039, over 14217.00 frames. ], tot_loss[loss=0.1431, simple_loss=0.1675, pruned_loss=0.05938, over 1938517.44 frames. ], batch size: 35, lr: 8.92e-03, grad_scale: 8.0 2022-12-07 20:00:07,219 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:00:14,337 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:00:23,176 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.029e+02 2.536e+02 3.250e+02 8.518e+02, threshold=5.072e+02, percent-clipped=1.0 2022-12-07 20:00:42,105 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7189, 1.9704, 2.1030, 2.0779, 1.8294, 2.1107, 1.7914, 1.3085], device='cuda:1'), covar=tensor([0.1471, 0.0918, 0.0490, 0.0402, 0.1244, 0.0612, 0.1563, 0.2396], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0069, 0.0055, 0.0058, 0.0089, 0.0068, 0.0091, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 20:01:09,227 INFO [train.py:873] (1/4) Epoch 9, batch 2400, loss[loss=0.1323, simple_loss=0.1653, pruned_loss=0.04963, over 14287.00 frames. ], tot_loss[loss=0.1441, simple_loss=0.168, pruned_loss=0.06013, over 1934531.36 frames. ], batch size: 39, lr: 8.91e-03, grad_scale: 8.0 2022-12-07 20:01:09,322 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2957, 3.0999, 2.8215, 2.9764, 3.2102, 3.1935, 3.2938, 3.2398], device='cuda:1'), covar=tensor([0.1011, 0.0667, 0.2111, 0.2879, 0.0780, 0.0965, 0.1141, 0.1051], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0244, 0.0411, 0.0521, 0.0301, 0.0386, 0.0378, 0.0340], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:01:12,682 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:14,007 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2022-12-07 20:01:21,566 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-07 20:01:34,484 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7125, 0.6828, 0.5767, 0.6753, 0.6774, 0.3422, 0.5358, 0.6421], device='cuda:1'), covar=tensor([0.0121, 0.0092, 0.0112, 0.0104, 0.0136, 0.0269, 0.0152, 0.0243], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.4709e-05, 1.0132e-04, 8.9602e-05, 9.8788e-05, 9.3031e-05, 1.4039e-04, 1.1894e-04, 1.3477e-04], device='cuda:1') 2022-12-07 20:01:36,917 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:47,457 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:49,793 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.251e+02 2.910e+02 3.795e+02 1.080e+03, threshold=5.819e+02, percent-clipped=6.0 2022-12-07 20:02:18,750 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:02:19,125 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 20:02:35,717 INFO [train.py:873] (1/4) Epoch 9, batch 2500, loss[loss=0.1586, simple_loss=0.1797, pruned_loss=0.06872, over 14420.00 frames. ], tot_loss[loss=0.1435, simple_loss=0.1681, pruned_loss=0.05941, over 2039270.99 frames. ], batch size: 73, lr: 8.90e-03, grad_scale: 8.0 2022-12-07 20:02:58,104 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:03:00,939 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4779, 1.3534, 1.4550, 1.2704, 1.3223, 0.9857, 1.0235, 1.0102], device='cuda:1'), covar=tensor([0.0135, 0.0283, 0.0285, 0.0268, 0.0182, 0.0319, 0.0209, 0.0333], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:1'), out_proj_covar=tensor([9.3242e-05, 1.0086e-04, 8.8605e-05, 9.8214e-05, 9.2477e-05, 1.3973e-04, 1.1827e-04, 1.3348e-04], device='cuda:1') 2022-12-07 20:03:17,451 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.289e+02 2.848e+02 3.593e+02 5.714e+02, threshold=5.696e+02, percent-clipped=0.0 2022-12-07 20:03:20,559 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:04:02,985 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:04:04,717 INFO [train.py:873] (1/4) Epoch 9, batch 2600, loss[loss=0.1211, simple_loss=0.1543, pruned_loss=0.04401, over 13959.00 frames. ], tot_loss[loss=0.1446, simple_loss=0.1683, pruned_loss=0.0605, over 1909733.03 frames. ], batch size: 26, lr: 8.90e-03, grad_scale: 8.0 2022-12-07 20:04:05,634 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5425, 1.2702, 1.5201, 2.0113, 1.5404, 1.4707, 1.5295, 1.7216], device='cuda:1'), covar=tensor([0.0835, 0.0874, 0.0895, 0.0518, 0.0555, 0.0522, 0.0544, 0.0802], device='cuda:1'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:1'), out_proj_covar=tensor([9.3265e-05, 1.0038e-04, 8.7965e-05, 9.7434e-05, 9.2386e-05, 1.3897e-04, 1.1776e-04, 1.3355e-04], device='cuda:1') 2022-12-07 20:04:32,672 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:04:45,881 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.604e+02 3.223e+02 3.963e+02 1.117e+03, threshold=6.445e+02, percent-clipped=7.0 2022-12-07 20:04:45,996 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0866, 2.0187, 1.6990, 1.8070, 2.0176, 2.0657, 2.0735, 2.0113], device='cuda:1'), covar=tensor([0.1250, 0.1152, 0.3168, 0.3179, 0.1274, 0.1271, 0.1618, 0.1388], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0244, 0.0407, 0.0520, 0.0297, 0.0382, 0.0375, 0.0337], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:05:25,932 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8885, 1.4401, 2.7257, 2.4690, 2.6540, 2.6685, 1.9691, 2.6846], device='cuda:1'), covar=tensor([0.0898, 0.1104, 0.0124, 0.0356, 0.0280, 0.0132, 0.0450, 0.0192], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0154, 0.0116, 0.0158, 0.0135, 0.0130, 0.0111, 0.0114], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:05:32,671 INFO [train.py:873] (1/4) Epoch 9, batch 2700, loss[loss=0.147, simple_loss=0.1721, pruned_loss=0.06093, over 12710.00 frames. ], tot_loss[loss=0.145, simple_loss=0.1685, pruned_loss=0.06072, over 1928585.53 frames. ], batch size: 100, lr: 8.89e-03, grad_scale: 8.0 2022-12-07 20:05:36,675 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:05:54,616 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:11,837 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:14,513 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.383e+02 3.002e+02 3.565e+02 6.845e+02, threshold=6.004e+02, percent-clipped=1.0 2022-12-07 20:06:16,777 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4275, 1.8569, 2.4574, 2.0689, 2.5620, 2.3294, 2.2083, 2.2006], device='cuda:1'), covar=tensor([0.0380, 0.2194, 0.0538, 0.1099, 0.0396, 0.0818, 0.0452, 0.1105], device='cuda:1'), in_proj_covar=tensor([0.0333, 0.0320, 0.0395, 0.0308, 0.0374, 0.0309, 0.0359, 0.0325], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:06:19,162 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:47,886 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:54,041 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:07:00,712 INFO [train.py:873] (1/4) Epoch 9, batch 2800, loss[loss=0.1531, simple_loss=0.1766, pruned_loss=0.06483, over 12739.00 frames. ], tot_loss[loss=0.1447, simple_loss=0.1686, pruned_loss=0.06039, over 1957864.82 frames. ], batch size: 100, lr: 8.88e-03, grad_scale: 8.0 2022-12-07 20:07:22,875 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:07:42,345 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.099e+02 2.541e+02 3.230e+02 5.695e+02, threshold=5.082e+02, percent-clipped=0.0 2022-12-07 20:07:48,598 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8798, 1.9404, 3.9684, 2.6548, 3.7419, 1.8535, 2.8926, 3.7230], device='cuda:1'), covar=tensor([0.0543, 0.4370, 0.0421, 0.6333, 0.0524, 0.3646, 0.1289, 0.0439], device='cuda:1'), in_proj_covar=tensor([0.0235, 0.0226, 0.0190, 0.0307, 0.0212, 0.0230, 0.0217, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:08:04,956 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:20,422 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3873, 1.3474, 1.3252, 1.3415, 1.4381, 0.7535, 1.2807, 1.4812], device='cuda:1'), covar=tensor([0.0873, 0.0910, 0.1260, 0.0634, 0.1225, 0.0700, 0.0611, 0.0838], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0022, 0.0022, 0.0021, 0.0023, 0.0032, 0.0022, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:1') 2022-12-07 20:08:27,862 INFO [train.py:873] (1/4) Epoch 9, batch 2900, loss[loss=0.1255, simple_loss=0.1488, pruned_loss=0.05112, over 4960.00 frames. ], tot_loss[loss=0.1437, simple_loss=0.1677, pruned_loss=0.05982, over 1936300.25 frames. ], batch size: 100, lr: 8.88e-03, grad_scale: 8.0 2022-12-07 20:08:48,393 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:50,034 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:55,935 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:01,096 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5925, 1.6511, 2.8500, 2.1437, 2.6997, 1.7176, 2.2810, 2.4730], device='cuda:1'), covar=tensor([0.1213, 0.4320, 0.0623, 0.4857, 0.0700, 0.3605, 0.1184, 0.1143], device='cuda:1'), in_proj_covar=tensor([0.0235, 0.0227, 0.0190, 0.0307, 0.0210, 0.0230, 0.0217, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:09:09,073 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.350e+02 2.806e+02 3.593e+02 6.969e+02, threshold=5.612e+02, percent-clipped=3.0 2022-12-07 20:09:24,865 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2126, 2.9794, 2.7473, 2.8998, 3.1141, 3.1553, 3.2059, 3.1681], device='cuda:1'), covar=tensor([0.0911, 0.0740, 0.2132, 0.2636, 0.0844, 0.0791, 0.1037, 0.0833], device='cuda:1'), in_proj_covar=tensor([0.0358, 0.0243, 0.0408, 0.0523, 0.0300, 0.0390, 0.0382, 0.0341], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:09:37,913 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:41,721 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:43,397 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:55,508 INFO [train.py:873] (1/4) Epoch 9, batch 3000, loss[loss=0.1333, simple_loss=0.1664, pruned_loss=0.05012, over 14269.00 frames. ], tot_loss[loss=0.1432, simple_loss=0.1678, pruned_loss=0.05932, over 1994051.83 frames. ], batch size: 60, lr: 8.87e-03, grad_scale: 8.0 2022-12-07 20:09:55,508 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 20:10:05,838 INFO [train.py:905] (1/4) Epoch 9, validation: loss=0.124, simple_loss=0.1667, pruned_loss=0.04063, over 857387.00 frames. 2022-12-07 20:10:05,839 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 20:10:30,008 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2022-12-07 20:10:38,676 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2158, 1.6819, 2.4566, 2.0516, 2.3580, 1.5983, 1.9064, 2.1946], device='cuda:1'), covar=tensor([0.1322, 0.3177, 0.0361, 0.2714, 0.0826, 0.2448, 0.1160, 0.0597], device='cuda:1'), in_proj_covar=tensor([0.0237, 0.0227, 0.0191, 0.0308, 0.0210, 0.0232, 0.0219, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:10:46,173 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.483e+02 2.932e+02 3.564e+02 6.010e+02, threshold=5.864e+02, percent-clipped=1.0 2022-12-07 20:11:00,656 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3789, 3.1499, 2.9539, 3.1048, 3.3432, 3.3013, 3.4081, 3.3586], device='cuda:1'), covar=tensor([0.1126, 0.0877, 0.2217, 0.2907, 0.0785, 0.0965, 0.1194, 0.0927], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0240, 0.0403, 0.0515, 0.0296, 0.0387, 0.0377, 0.0336], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:11:08,809 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:15,605 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:23,895 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:33,371 INFO [train.py:873] (1/4) Epoch 9, batch 3100, loss[loss=0.1756, simple_loss=0.1673, pruned_loss=0.09198, over 2643.00 frames. ], tot_loss[loss=0.1439, simple_loss=0.1678, pruned_loss=0.06003, over 1979372.67 frames. ], batch size: 100, lr: 8.86e-03, grad_scale: 8.0 2022-12-07 20:12:03,492 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:12:15,373 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.304e+02 2.853e+02 3.296e+02 8.516e+02, threshold=5.706e+02, percent-clipped=4.0 2022-12-07 20:12:18,595 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:13:02,478 INFO [train.py:873] (1/4) Epoch 9, batch 3200, loss[loss=0.1499, simple_loss=0.1653, pruned_loss=0.06718, over 6969.00 frames. ], tot_loss[loss=0.1429, simple_loss=0.1671, pruned_loss=0.05941, over 1944698.65 frames. ], batch size: 100, lr: 8.86e-03, grad_scale: 8.0 2022-12-07 20:13:09,755 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0929, 3.5396, 2.9335, 4.3555, 4.1747, 4.2217, 3.6161, 3.0388], device='cuda:1'), covar=tensor([0.0973, 0.1477, 0.3897, 0.0513, 0.0771, 0.1484, 0.1129, 0.3572], device='cuda:1'), in_proj_covar=tensor([0.0260, 0.0299, 0.0277, 0.0236, 0.0294, 0.0287, 0.0257, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 20:13:29,563 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4337, 3.4170, 3.1744, 3.2820, 3.3107, 3.3562, 1.3521, 3.1654], device='cuda:1'), covar=tensor([0.0507, 0.0534, 0.1117, 0.0737, 0.0721, 0.0629, 0.4615, 0.0636], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0131, 0.0130, 0.0188, 0.0129, 0.0156, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:13:34,872 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-07 20:13:43,328 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 2.396e+02 2.898e+02 3.837e+02 7.541e+02, threshold=5.796e+02, percent-clipped=6.0 2022-12-07 20:14:11,316 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:14:13,043 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:14:15,964 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.09 vs. limit=2.0 2022-12-07 20:14:29,969 INFO [train.py:873] (1/4) Epoch 9, batch 3300, loss[loss=0.117, simple_loss=0.1562, pruned_loss=0.03892, over 14321.00 frames. ], tot_loss[loss=0.1438, simple_loss=0.1675, pruned_loss=0.06009, over 1913844.13 frames. ], batch size: 25, lr: 8.85e-03, grad_scale: 8.0 2022-12-07 20:14:41,709 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:14:56,616 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1649, 1.5465, 1.6902, 1.7010, 1.5510, 1.7477, 1.3501, 1.2422], device='cuda:1'), covar=tensor([0.1673, 0.1103, 0.0382, 0.0344, 0.0965, 0.0553, 0.1958, 0.1656], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0070, 0.0055, 0.0058, 0.0087, 0.0067, 0.0093, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 20:15:10,959 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.144e+02 2.539e+02 3.073e+02 7.001e+02, threshold=5.078e+02, percent-clipped=2.0 2022-12-07 20:15:12,061 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:34,967 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:40,152 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:56,935 INFO [train.py:873] (1/4) Epoch 9, batch 3400, loss[loss=0.133, simple_loss=0.1572, pruned_loss=0.05439, over 14144.00 frames. ], tot_loss[loss=0.1418, simple_loss=0.1663, pruned_loss=0.05864, over 1931635.69 frames. ], batch size: 25, lr: 8.84e-03, grad_scale: 8.0 2022-12-07 20:16:05,379 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:14,126 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:21,650 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:21,684 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:36,645 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:38,551 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.128e+02 2.612e+02 3.647e+02 5.829e+02, threshold=5.224e+02, percent-clipped=2.0 2022-12-07 20:16:56,868 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3123, 3.9804, 3.9485, 4.3075, 4.0775, 3.7242, 4.3161, 3.6149], device='cuda:1'), covar=tensor([0.0465, 0.1000, 0.0377, 0.0493, 0.0890, 0.1104, 0.0618, 0.0582], device='cuda:1'), in_proj_covar=tensor([0.0161, 0.0251, 0.0173, 0.0168, 0.0168, 0.0135, 0.0256, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 20:17:07,541 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:17:25,060 INFO [train.py:873] (1/4) Epoch 9, batch 3500, loss[loss=0.1515, simple_loss=0.1733, pruned_loss=0.06487, over 14364.00 frames. ], tot_loss[loss=0.1417, simple_loss=0.1664, pruned_loss=0.05854, over 1945259.31 frames. ], batch size: 73, lr: 8.83e-03, grad_scale: 8.0 2022-12-07 20:17:48,241 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:06,279 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.173e+02 2.574e+02 3.096e+02 7.573e+02, threshold=5.148e+02, percent-clipped=3.0 2022-12-07 20:18:20,201 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0796, 1.2351, 3.2424, 1.4295, 3.0871, 3.1730, 2.1611, 3.3814], device='cuda:1'), covar=tensor([0.0275, 0.3135, 0.0335, 0.2356, 0.0819, 0.0442, 0.1089, 0.0214], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0155, 0.0154, 0.0169, 0.0166, 0.0164, 0.0134, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:18:33,628 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:35,424 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:40,787 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:51,457 INFO [train.py:873] (1/4) Epoch 9, batch 3600, loss[loss=0.1234, simple_loss=0.1651, pruned_loss=0.04084, over 14409.00 frames. ], tot_loss[loss=0.141, simple_loss=0.1664, pruned_loss=0.05784, over 1962131.50 frames. ], batch size: 41, lr: 8.83e-03, grad_scale: 8.0 2022-12-07 20:19:15,031 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:19:16,695 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:19:32,556 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.135e+01 2.573e+02 3.171e+02 4.040e+02 8.310e+02, threshold=6.341e+02, percent-clipped=10.0 2022-12-07 20:19:42,859 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 20:19:51,905 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:05,062 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9819, 1.4443, 3.1351, 1.5678, 3.2370, 3.1531, 2.3311, 3.2794], device='cuda:1'), covar=tensor([0.0243, 0.2653, 0.0308, 0.1933, 0.0303, 0.0377, 0.0805, 0.0195], device='cuda:1'), in_proj_covar=tensor([0.0167, 0.0156, 0.0156, 0.0170, 0.0167, 0.0165, 0.0134, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:20:06,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2022-12-07 20:20:12,687 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0787, 1.1447, 0.9642, 1.0057, 1.2253, 0.4373, 0.8932, 0.9015], device='cuda:1'), covar=tensor([0.0657, 0.0719, 0.1035, 0.0541, 0.0521, 0.0713, 0.0841, 0.0568], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0023, 0.0023, 0.0023, 0.0024, 0.0034, 0.0023, 0.0025], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 20:20:18,991 INFO [train.py:873] (1/4) Epoch 9, batch 3700, loss[loss=0.14, simple_loss=0.1664, pruned_loss=0.05675, over 13539.00 frames. ], tot_loss[loss=0.1416, simple_loss=0.1671, pruned_loss=0.05805, over 2040252.44 frames. ], batch size: 100, lr: 8.82e-03, grad_scale: 8.0 2022-12-07 20:20:22,587 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:32,614 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2152, 1.8101, 4.5959, 4.1864, 4.1077, 4.6746, 4.2276, 4.6369], device='cuda:1'), covar=tensor([0.1231, 0.1490, 0.0101, 0.0161, 0.0169, 0.0128, 0.0121, 0.0134], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0119, 0.0161, 0.0138, 0.0134, 0.0113, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:20:44,474 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:50,631 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:59,003 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:00,574 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.090e+02 2.323e+02 2.858e+02 3.638e+02 5.583e+02, threshold=5.716e+02, percent-clipped=0.0 2022-12-07 20:21:25,187 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:25,998 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:41,392 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:44,065 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:47,276 INFO [train.py:873] (1/4) Epoch 9, batch 3800, loss[loss=0.1532, simple_loss=0.1497, pruned_loss=0.07832, over 3883.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1674, pruned_loss=0.05864, over 2035385.76 frames. ], batch size: 100, lr: 8.81e-03, grad_scale: 8.0 2022-12-07 20:22:28,598 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.269e+02 2.644e+02 3.472e+02 7.922e+02, threshold=5.288e+02, percent-clipped=2.0 2022-12-07 20:22:29,737 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:22:59,840 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:23:16,036 INFO [train.py:873] (1/4) Epoch 9, batch 3900, loss[loss=0.1676, simple_loss=0.156, pruned_loss=0.0896, over 2648.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1671, pruned_loss=0.05875, over 2040591.87 frames. ], batch size: 100, lr: 8.81e-03, grad_scale: 8.0 2022-12-07 20:23:24,320 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:23:57,837 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.154e+02 2.432e+02 3.241e+02 6.644e+02, threshold=4.864e+02, percent-clipped=3.0 2022-12-07 20:24:17,237 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8123, 5.1999, 5.1921, 5.7308, 5.3946, 4.7236, 5.6175, 4.6093], device='cuda:1'), covar=tensor([0.0272, 0.1019, 0.0292, 0.0363, 0.0685, 0.0369, 0.0562, 0.0537], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0250, 0.0173, 0.0165, 0.0167, 0.0135, 0.0252, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 20:24:17,350 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:24:18,142 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:24:31,973 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4371, 2.4195, 4.6375, 2.8546, 4.1403, 1.8509, 3.1449, 4.0821], device='cuda:1'), covar=tensor([0.0686, 0.4636, 0.0366, 0.8575, 0.0777, 0.4986, 0.1540, 0.0503], device='cuda:1'), in_proj_covar=tensor([0.0237, 0.0225, 0.0193, 0.0305, 0.0212, 0.0229, 0.0219, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:24:45,107 INFO [train.py:873] (1/4) Epoch 9, batch 4000, loss[loss=0.1109, simple_loss=0.1472, pruned_loss=0.03735, over 13929.00 frames. ], tot_loss[loss=0.1422, simple_loss=0.167, pruned_loss=0.05871, over 2032340.46 frames. ], batch size: 26, lr: 8.80e-03, grad_scale: 8.0 2022-12-07 20:24:48,708 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:00,316 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:11,113 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:26,920 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.384e+02 2.847e+02 3.821e+02 7.495e+02, threshold=5.695e+02, percent-clipped=8.0 2022-12-07 20:25:31,513 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:51,471 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:06,385 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:07,633 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:12,705 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3878, 1.9823, 2.7750, 2.1534, 1.8159, 2.4206, 1.1986, 2.4163], device='cuda:1'), covar=tensor([0.1861, 0.2562, 0.0976, 0.2045, 0.3423, 0.1319, 0.5903, 0.1376], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0091, 0.0085, 0.0090, 0.0108, 0.0078, 0.0125, 0.0082], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:26:14,196 INFO [train.py:873] (1/4) Epoch 9, batch 4100, loss[loss=0.1639, simple_loss=0.179, pruned_loss=0.07444, over 14254.00 frames. ], tot_loss[loss=0.1426, simple_loss=0.1671, pruned_loss=0.05899, over 1957637.31 frames. ], batch size: 76, lr: 8.79e-03, grad_scale: 8.0 2022-12-07 20:26:22,150 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2569, 2.2198, 3.1795, 2.4713, 3.2982, 3.1032, 3.0934, 2.5502], device='cuda:1'), covar=tensor([0.0583, 0.3152, 0.0747, 0.2084, 0.0559, 0.0772, 0.1322, 0.2953], device='cuda:1'), in_proj_covar=tensor([0.0329, 0.0316, 0.0389, 0.0304, 0.0371, 0.0307, 0.0361, 0.0318], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:26:34,414 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:55,696 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.345e+01 2.214e+02 2.881e+02 3.617e+02 6.891e+02, threshold=5.761e+02, percent-clipped=1.0 2022-12-07 20:26:56,751 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0753, 4.8852, 4.4205, 4.6685, 4.7346, 4.9666, 5.1310, 5.1092], device='cuda:1'), covar=tensor([0.0939, 0.0473, 0.2643, 0.2983, 0.0881, 0.0852, 0.0907, 0.0815], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0239, 0.0405, 0.0512, 0.0295, 0.0385, 0.0374, 0.0336], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:26:59,475 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:01,254 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:11,203 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 20:27:26,446 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:41,954 INFO [train.py:873] (1/4) Epoch 9, batch 4200, loss[loss=0.1452, simple_loss=0.1629, pruned_loss=0.06373, over 9503.00 frames. ], tot_loss[loss=0.141, simple_loss=0.1664, pruned_loss=0.05776, over 1983023.27 frames. ], batch size: 100, lr: 8.79e-03, grad_scale: 8.0 2022-12-07 20:27:45,417 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:46,948 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:52,512 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:07,769 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:08,615 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7639, 3.6498, 3.5514, 3.8664, 3.4115, 3.3284, 3.8724, 3.7274], device='cuda:1'), covar=tensor([0.0710, 0.0739, 0.0746, 0.0589, 0.0904, 0.0653, 0.0631, 0.0818], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0120, 0.0132, 0.0136, 0.0132, 0.0108, 0.0150, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 20:28:11,685 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 20:28:23,777 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.280e+02 2.745e+02 3.613e+02 6.444e+02, threshold=5.491e+02, percent-clipped=3.0 2022-12-07 20:28:23,929 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:30,348 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:40,216 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:09,365 INFO [train.py:873] (1/4) Epoch 9, batch 4300, loss[loss=0.1329, simple_loss=0.1678, pruned_loss=0.04898, over 14395.00 frames. ], tot_loss[loss=0.1422, simple_loss=0.1676, pruned_loss=0.05843, over 1980023.00 frames. ], batch size: 41, lr: 8.78e-03, grad_scale: 8.0 2022-12-07 20:29:17,497 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:23,527 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:30,626 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:51,886 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.326e+02 2.910e+02 3.508e+02 7.763e+02, threshold=5.820e+02, percent-clipped=3.0 2022-12-07 20:30:01,862 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 20:30:29,589 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:30:37,025 INFO [train.py:873] (1/4) Epoch 9, batch 4400, loss[loss=0.1545, simple_loss=0.1633, pruned_loss=0.07281, over 4952.00 frames. ], tot_loss[loss=0.1429, simple_loss=0.1679, pruned_loss=0.05893, over 1982740.00 frames. ], batch size: 100, lr: 8.77e-03, grad_scale: 8.0 2022-12-07 20:31:10,777 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:31:18,744 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.162e+02 2.178e+02 2.781e+02 3.445e+02 8.866e+02, threshold=5.563e+02, percent-clipped=5.0 2022-12-07 20:31:18,895 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:31:20,073 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2022-12-07 20:31:45,921 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:04,726 INFO [train.py:873] (1/4) Epoch 9, batch 4500, loss[loss=0.1458, simple_loss=0.1418, pruned_loss=0.07485, over 2600.00 frames. ], tot_loss[loss=0.1415, simple_loss=0.1668, pruned_loss=0.05808, over 1943230.93 frames. ], batch size: 100, lr: 8.77e-03, grad_scale: 8.0 2022-12-07 20:32:11,661 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:14,103 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:17,621 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8503, 1.9423, 2.1010, 1.4128, 1.5499, 1.9705, 1.0790, 1.8282], device='cuda:1'), covar=tensor([0.1453, 0.2033, 0.0931, 0.2396, 0.2658, 0.1233, 0.4285, 0.1104], device='cuda:1'), in_proj_covar=tensor([0.0076, 0.0092, 0.0087, 0.0092, 0.0110, 0.0079, 0.0128, 0.0083], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:32:23,583 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4939, 3.3984, 3.2970, 3.5921, 3.2071, 2.9707, 3.5540, 3.4694], device='cuda:1'), covar=tensor([0.0712, 0.0816, 0.0800, 0.0634, 0.0965, 0.0729, 0.0825, 0.0739], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0120, 0.0132, 0.0137, 0.0132, 0.0108, 0.0150, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 20:32:43,544 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 20:32:45,991 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7073, 2.6001, 1.9634, 2.6684, 2.4803, 2.5776, 2.4018, 2.1551], device='cuda:1'), covar=tensor([0.0606, 0.1238, 0.2699, 0.0622, 0.1046, 0.0906, 0.1302, 0.2373], device='cuda:1'), in_proj_covar=tensor([0.0259, 0.0299, 0.0277, 0.0231, 0.0296, 0.0289, 0.0257, 0.0265], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 20:32:50,125 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.276e+02 2.761e+02 3.494e+02 6.954e+02, threshold=5.522e+02, percent-clipped=1.0 2022-12-07 20:32:53,585 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:53,680 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1771, 1.8217, 2.1819, 1.3539, 1.8719, 2.1166, 2.2418, 1.9620], device='cuda:1'), covar=tensor([0.0899, 0.0951, 0.1039, 0.1853, 0.1392, 0.0744, 0.0617, 0.1569], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0183, 0.0131, 0.0123, 0.0125, 0.0131, 0.0107, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:1') 2022-12-07 20:32:58,551 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:02,175 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:35,180 INFO [train.py:873] (1/4) Epoch 9, batch 4600, loss[loss=0.1379, simple_loss=0.1693, pruned_loss=0.05326, over 14278.00 frames. ], tot_loss[loss=0.1429, simple_loss=0.1679, pruned_loss=0.059, over 1932804.92 frames. ], batch size: 76, lr: 8.76e-03, grad_scale: 8.0 2022-12-07 20:33:36,762 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 20:33:39,019 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:45,172 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:52,358 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:56,528 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:34:17,217 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.207e+01 2.339e+02 3.104e+02 3.874e+02 1.034e+03, threshold=6.208e+02, percent-clipped=6.0 2022-12-07 20:34:38,794 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:34:56,827 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6696, 1.3787, 3.7260, 1.7463, 3.6737, 3.8720, 2.9539, 4.0232], device='cuda:1'), covar=tensor([0.0259, 0.3076, 0.0351, 0.2312, 0.0554, 0.0355, 0.0670, 0.0195], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0155, 0.0155, 0.0167, 0.0167, 0.0167, 0.0133, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:35:03,849 INFO [train.py:873] (1/4) Epoch 9, batch 4700, loss[loss=0.1314, simple_loss=0.1733, pruned_loss=0.0447, over 14242.00 frames. ], tot_loss[loss=0.1418, simple_loss=0.1672, pruned_loss=0.05824, over 2001422.41 frames. ], batch size: 35, lr: 8.75e-03, grad_scale: 8.0 2022-12-07 20:35:27,771 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 20:35:46,990 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.284e+02 2.894e+02 3.637e+02 5.922e+02, threshold=5.788e+02, percent-clipped=0.0 2022-12-07 20:35:47,122 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:21,946 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:28,430 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:31,781 INFO [train.py:873] (1/4) Epoch 9, batch 4800, loss[loss=0.1547, simple_loss=0.1477, pruned_loss=0.08084, over 2636.00 frames. ], tot_loss[loss=0.1417, simple_loss=0.1668, pruned_loss=0.05829, over 1942520.01 frames. ], batch size: 100, lr: 8.75e-03, grad_scale: 8.0 2022-12-07 20:36:38,344 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:02,592 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 20:37:13,781 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.162e+02 2.794e+02 3.355e+02 5.841e+02, threshold=5.587e+02, percent-clipped=1.0 2022-12-07 20:37:14,876 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:20,538 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:26,002 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:58,505 INFO [train.py:873] (1/4) Epoch 9, batch 4900, loss[loss=0.2038, simple_loss=0.1811, pruned_loss=0.1133, over 1300.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1671, pruned_loss=0.05871, over 1930661.85 frames. ], batch size: 100, lr: 8.74e-03, grad_scale: 8.0 2022-12-07 20:38:02,312 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:07,260 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:07,628 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2022-12-07 20:38:08,184 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:10,629 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:22,200 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7268, 1.7916, 1.8709, 1.3475, 1.3351, 1.7840, 1.1152, 1.6558], device='cuda:1'), covar=tensor([0.1052, 0.1786, 0.0859, 0.2525, 0.2517, 0.0748, 0.2904, 0.0828], device='cuda:1'), in_proj_covar=tensor([0.0075, 0.0092, 0.0086, 0.0092, 0.0110, 0.0078, 0.0128, 0.0083], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:38:40,335 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.955e+01 2.399e+02 2.803e+02 3.609e+02 7.825e+02, threshold=5.606e+02, percent-clipped=2.0 2022-12-07 20:38:43,761 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:49,505 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:39:25,266 INFO [train.py:873] (1/4) Epoch 9, batch 5000, loss[loss=0.1617, simple_loss=0.1697, pruned_loss=0.07688, over 4971.00 frames. ], tot_loss[loss=0.1419, simple_loss=0.1671, pruned_loss=0.05839, over 1911323.14 frames. ], batch size: 100, lr: 8.73e-03, grad_scale: 8.0 2022-12-07 20:39:56,492 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:40:07,679 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.681e+01 2.367e+02 2.989e+02 3.809e+02 1.039e+03, threshold=5.978e+02, percent-clipped=5.0 2022-12-07 20:40:16,249 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4013, 1.6105, 1.5987, 1.4541, 1.3313, 1.2686, 0.9944, 0.9867], device='cuda:1'), covar=tensor([0.0253, 0.0309, 0.0171, 0.0150, 0.0229, 0.0222, 0.0185, 0.0286], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0013, 0.0020, 0.0016, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.6829e-05, 1.0514e-04, 9.2712e-05, 9.9533e-05, 9.7965e-05, 1.4410e-04, 1.2035e-04, 1.4034e-04], device='cuda:1') 2022-12-07 20:40:50,492 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:40:53,898 INFO [train.py:873] (1/4) Epoch 9, batch 5100, loss[loss=0.1244, simple_loss=0.1582, pruned_loss=0.04531, over 14371.00 frames. ], tot_loss[loss=0.1413, simple_loss=0.1665, pruned_loss=0.05802, over 1960033.07 frames. ], batch size: 18, lr: 8.73e-03, grad_scale: 8.0 2022-12-07 20:41:14,430 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 20:41:23,438 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0913, 2.1262, 4.9163, 4.5567, 4.3243, 5.0148, 4.8331, 5.0585], device='cuda:1'), covar=tensor([0.1322, 0.1271, 0.0076, 0.0143, 0.0157, 0.0084, 0.0077, 0.0080], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0121, 0.0164, 0.0139, 0.0135, 0.0114, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:41:25,186 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:41:33,035 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:41:36,452 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.306e+02 2.958e+02 3.852e+02 6.809e+02, threshold=5.917e+02, percent-clipped=2.0 2022-12-07 20:42:07,894 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:42:20,479 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 20:42:22,489 INFO [train.py:873] (1/4) Epoch 9, batch 5200, loss[loss=0.1588, simple_loss=0.1751, pruned_loss=0.0712, over 7726.00 frames. ], tot_loss[loss=0.1419, simple_loss=0.167, pruned_loss=0.05834, over 1990036.44 frames. ], batch size: 100, lr: 8.72e-03, grad_scale: 8.0 2022-12-07 20:42:35,340 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:43:04,806 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.172e+02 3.153e+02 3.803e+02 8.139e+02, threshold=6.306e+02, percent-clipped=3.0 2022-12-07 20:43:17,696 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:43:39,119 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8444, 0.8381, 0.6818, 0.6800, 0.6148, 0.6205, 0.5818, 0.6171], device='cuda:1'), covar=tensor([0.0127, 0.0157, 0.0151, 0.0150, 0.0245, 0.0322, 0.0223, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.7886e-05, 1.0671e-04, 9.3727e-05, 1.0102e-04, 9.9288e-05, 1.4735e-04, 1.2321e-04, 1.4270e-04], device='cuda:1') 2022-12-07 20:43:43,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8944, 1.2759, 2.0583, 1.1562, 2.0304, 2.0634, 1.6292, 2.1133], device='cuda:1'), covar=tensor([0.0292, 0.1662, 0.0321, 0.1524, 0.0413, 0.0528, 0.0910, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0159, 0.0158, 0.0170, 0.0168, 0.0172, 0.0137, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:43:50,698 INFO [train.py:873] (1/4) Epoch 9, batch 5300, loss[loss=0.1184, simple_loss=0.1482, pruned_loss=0.0443, over 13912.00 frames. ], tot_loss[loss=0.1412, simple_loss=0.1664, pruned_loss=0.058, over 1908790.63 frames. ], batch size: 20, lr: 8.71e-03, grad_scale: 8.0 2022-12-07 20:44:26,680 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:44:32,252 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.588e+01 2.128e+02 2.955e+02 3.598e+02 1.062e+03, threshold=5.909e+02, percent-clipped=4.0 2022-12-07 20:44:44,221 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7231, 3.5242, 3.5731, 3.8769, 3.4389, 3.1973, 3.8971, 3.7366], device='cuda:1'), covar=tensor([0.0775, 0.1048, 0.0861, 0.0632, 0.0922, 0.0757, 0.0669, 0.0820], device='cuda:1'), in_proj_covar=tensor([0.0127, 0.0119, 0.0131, 0.0137, 0.0132, 0.0107, 0.0150, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 20:44:47,065 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6368, 1.4335, 4.4044, 1.8345, 4.2288, 4.3661, 4.1283, 4.9154], device='cuda:1'), covar=tensor([0.0192, 0.3110, 0.0333, 0.2283, 0.0347, 0.0483, 0.0382, 0.0160], device='cuda:1'), in_proj_covar=tensor([0.0169, 0.0157, 0.0156, 0.0169, 0.0168, 0.0171, 0.0136, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 20:44:53,141 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9383, 1.8203, 3.2161, 2.2062, 2.8985, 1.7401, 2.4366, 2.9081], device='cuda:1'), covar=tensor([0.0836, 0.4270, 0.0450, 0.4946, 0.0778, 0.3600, 0.1293, 0.0654], device='cuda:1'), in_proj_covar=tensor([0.0241, 0.0224, 0.0195, 0.0304, 0.0214, 0.0232, 0.0220, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:45:09,732 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:45:17,381 INFO [train.py:873] (1/4) Epoch 9, batch 5400, loss[loss=0.1595, simple_loss=0.1648, pruned_loss=0.07711, over 3876.00 frames. ], tot_loss[loss=0.1408, simple_loss=0.1662, pruned_loss=0.0577, over 1973694.33 frames. ], batch size: 100, lr: 8.71e-03, grad_scale: 8.0 2022-12-07 20:45:19,407 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:45:49,528 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1815, 1.3744, 1.2120, 1.0603, 0.8878, 0.9529, 0.8350, 0.7946], device='cuda:1'), covar=tensor([0.0198, 0.0312, 0.0208, 0.0285, 0.0273, 0.0448, 0.0254, 0.0482], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0021], device='cuda:1'), out_proj_covar=tensor([9.6873e-05, 1.0566e-04, 9.3020e-05, 1.0031e-04, 9.7646e-05, 1.4637e-04, 1.2194e-04, 1.4169e-04], device='cuda:1') 2022-12-07 20:45:56,528 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:46:00,157 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.009e+02 2.028e+02 2.560e+02 3.349e+02 6.726e+02, threshold=5.119e+02, percent-clipped=1.0 2022-12-07 20:46:38,486 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:46:45,301 INFO [train.py:873] (1/4) Epoch 9, batch 5500, loss[loss=0.1098, simple_loss=0.1463, pruned_loss=0.03668, over 14058.00 frames. ], tot_loss[loss=0.1398, simple_loss=0.1654, pruned_loss=0.0571, over 1956585.94 frames. ], batch size: 22, lr: 8.70e-03, grad_scale: 8.0 2022-12-07 20:47:00,124 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 20:47:27,195 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.224e+02 2.797e+02 3.647e+02 1.046e+03, threshold=5.594e+02, percent-clipped=7.0 2022-12-07 20:48:11,086 INFO [train.py:873] (1/4) Epoch 9, batch 5600, loss[loss=0.1239, simple_loss=0.1594, pruned_loss=0.04423, over 14532.00 frames. ], tot_loss[loss=0.1416, simple_loss=0.1667, pruned_loss=0.05826, over 1967302.41 frames. ], batch size: 43, lr: 8.69e-03, grad_scale: 8.0 2022-12-07 20:48:52,367 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:48:53,077 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.262e+02 2.866e+02 3.556e+02 6.245e+02, threshold=5.731e+02, percent-clipped=4.0 2022-12-07 20:49:07,535 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:49:32,004 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:49:37,415 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-07 20:49:37,631 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:49:40,171 INFO [train.py:873] (1/4) Epoch 9, batch 5700, loss[loss=0.1415, simple_loss=0.1738, pruned_loss=0.05465, over 14114.00 frames. ], tot_loss[loss=0.142, simple_loss=0.1669, pruned_loss=0.05852, over 2014632.30 frames. ], batch size: 29, lr: 8.69e-03, grad_scale: 8.0 2022-12-07 20:49:47,579 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 20:50:01,201 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 20:50:15,198 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:50:17,860 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8671, 1.6332, 2.1210, 1.5880, 1.9493, 1.5036, 1.7562, 1.7996], device='cuda:1'), covar=tensor([0.1903, 0.1825, 0.0284, 0.1109, 0.1038, 0.1301, 0.0737, 0.0706], device='cuda:1'), in_proj_covar=tensor([0.0240, 0.0224, 0.0192, 0.0303, 0.0211, 0.0232, 0.0220, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:50:23,973 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.292e+02 2.851e+02 3.352e+02 5.991e+02, threshold=5.702e+02, percent-clipped=1.0 2022-12-07 20:50:26,792 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2320, 1.0326, 1.0734, 1.1465, 1.3259, 0.7935, 1.0418, 1.2077], device='cuda:1'), covar=tensor([0.0577, 0.0939, 0.0873, 0.1036, 0.0683, 0.0762, 0.0743, 0.0553], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0023, 0.0025, 0.0023, 0.0024, 0.0034, 0.0024, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 20:50:34,244 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:51:08,643 INFO [train.py:873] (1/4) Epoch 9, batch 5800, loss[loss=0.1504, simple_loss=0.1771, pruned_loss=0.06188, over 14289.00 frames. ], tot_loss[loss=0.1431, simple_loss=0.1679, pruned_loss=0.05919, over 2036512.33 frames. ], batch size: 25, lr: 8.68e-03, grad_scale: 4.0 2022-12-07 20:51:28,515 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:51:51,562 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.104e+02 2.481e+02 2.990e+02 4.125e+02 8.241e+02, threshold=5.981e+02, percent-clipped=12.0 2022-12-07 20:52:02,551 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2022-12-07 20:52:36,170 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7539, 1.2142, 1.3521, 1.2828, 1.1125, 1.3613, 1.0340, 0.8416], device='cuda:1'), covar=tensor([0.2261, 0.0669, 0.0275, 0.0264, 0.1144, 0.0565, 0.1583, 0.1040], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0071, 0.0056, 0.0058, 0.0088, 0.0067, 0.0093, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 20:52:36,916 INFO [train.py:873] (1/4) Epoch 9, batch 5900, loss[loss=0.192, simple_loss=0.1638, pruned_loss=0.1101, over 1258.00 frames. ], tot_loss[loss=0.1424, simple_loss=0.1675, pruned_loss=0.05867, over 2028600.61 frames. ], batch size: 100, lr: 8.67e-03, grad_scale: 4.0 2022-12-07 20:52:47,026 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2022-12-07 20:52:51,572 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-07 20:53:12,734 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-07 20:53:19,921 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.452e+02 3.007e+02 3.915e+02 6.171e+02, threshold=6.013e+02, percent-clipped=2.0 2022-12-07 20:53:26,059 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4168, 1.0609, 1.3230, 1.3433, 1.5585, 0.6887, 1.3013, 1.2898], device='cuda:1'), covar=tensor([0.0610, 0.1053, 0.0656, 0.0977, 0.0798, 0.0690, 0.0622, 0.0456], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0023, 0.0025, 0.0022, 0.0023, 0.0033, 0.0023, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 20:54:02,071 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:54:04,622 INFO [train.py:873] (1/4) Epoch 9, batch 6000, loss[loss=0.1192, simple_loss=0.1565, pruned_loss=0.04094, over 14080.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1672, pruned_loss=0.05873, over 1998342.97 frames. ], batch size: 22, lr: 8.67e-03, grad_scale: 8.0 2022-12-07 20:54:04,622 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 20:54:16,409 INFO [train.py:905] (1/4) Epoch 9, validation: loss=0.1244, simple_loss=0.166, pruned_loss=0.04137, over 857387.00 frames. 2022-12-07 20:54:16,410 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 20:54:19,133 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 20:54:32,906 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:54:37,370 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:54:44,424 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 20:54:56,078 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:54:59,452 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.842e+01 2.449e+02 2.904e+02 3.531e+02 7.481e+02, threshold=5.809e+02, percent-clipped=4.0 2022-12-07 20:55:31,917 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:55:45,073 INFO [train.py:873] (1/4) Epoch 9, batch 6100, loss[loss=0.1608, simple_loss=0.1805, pruned_loss=0.07049, over 13530.00 frames. ], tot_loss[loss=0.142, simple_loss=0.1669, pruned_loss=0.05859, over 1974634.26 frames. ], batch size: 100, lr: 8.66e-03, grad_scale: 8.0 2022-12-07 20:55:50,406 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9848, 2.0845, 1.8867, 2.1248, 1.7070, 2.0099, 2.0702, 2.0526], device='cuda:1'), covar=tensor([0.0929, 0.1005, 0.1223, 0.0827, 0.1416, 0.0978, 0.1188, 0.0905], device='cuda:1'), in_proj_covar=tensor([0.0128, 0.0120, 0.0132, 0.0139, 0.0132, 0.0108, 0.0152, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 20:55:58,512 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2185, 2.0143, 4.7930, 4.3195, 4.3345, 4.9083, 4.6830, 4.9655], device='cuda:1'), covar=tensor([0.1328, 0.1418, 0.0077, 0.0168, 0.0158, 0.0080, 0.0074, 0.0091], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0122, 0.0166, 0.0139, 0.0136, 0.0114, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:56:00,273 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:56:15,649 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2022-12-07 20:56:28,746 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.301e+02 2.963e+02 3.616e+02 7.389e+02, threshold=5.926e+02, percent-clipped=2.0 2022-12-07 20:56:54,812 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6841, 3.7616, 3.9374, 3.5758, 3.8147, 3.8111, 1.5221, 3.5763], device='cuda:1'), covar=tensor([0.0292, 0.0367, 0.0358, 0.0454, 0.0320, 0.0326, 0.3043, 0.0284], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0157, 0.0130, 0.0129, 0.0186, 0.0127, 0.0152, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 20:57:13,996 INFO [train.py:873] (1/4) Epoch 9, batch 6200, loss[loss=0.1327, simple_loss=0.1694, pruned_loss=0.04803, over 14307.00 frames. ], tot_loss[loss=0.1424, simple_loss=0.167, pruned_loss=0.05887, over 1953399.36 frames. ], batch size: 35, lr: 8.66e-03, grad_scale: 8.0 2022-12-07 20:57:44,154 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:57:51,413 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2084, 3.9152, 3.6222, 3.7721, 4.0042, 4.0928, 4.1910, 4.1772], device='cuda:1'), covar=tensor([0.0852, 0.0514, 0.1981, 0.2981, 0.0640, 0.0773, 0.0949, 0.0851], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0244, 0.0409, 0.0526, 0.0305, 0.0390, 0.0380, 0.0345], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 20:57:57,368 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.480e+01 2.278e+02 2.813e+02 3.570e+02 7.547e+02, threshold=5.626e+02, percent-clipped=4.0 2022-12-07 20:58:11,611 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 20:58:24,865 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-07 20:58:37,967 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:58:42,095 INFO [train.py:873] (1/4) Epoch 9, batch 6300, loss[loss=0.1171, simple_loss=0.1553, pruned_loss=0.03938, over 14222.00 frames. ], tot_loss[loss=0.1415, simple_loss=0.1668, pruned_loss=0.05806, over 2009705.29 frames. ], batch size: 37, lr: 8.65e-03, grad_scale: 8.0 2022-12-07 20:58:45,105 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:58:59,261 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:59:08,871 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.59 vs. limit=5.0 2022-12-07 20:59:25,549 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.076e+02 2.178e+02 2.874e+02 3.614e+02 8.081e+02, threshold=5.749e+02, percent-clipped=6.0 2022-12-07 20:59:27,371 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:59:35,404 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5727, 3.0167, 2.6801, 2.8044, 2.1484, 2.8906, 2.7251, 1.1325], device='cuda:1'), covar=tensor([0.2506, 0.0729, 0.1478, 0.0892, 0.1184, 0.0565, 0.1246, 0.3156], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0070, 0.0057, 0.0058, 0.0086, 0.0066, 0.0091, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 20:59:42,255 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:59:53,363 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:00:10,824 INFO [train.py:873] (1/4) Epoch 9, batch 6400, loss[loss=0.1516, simple_loss=0.172, pruned_loss=0.0656, over 14232.00 frames. ], tot_loss[loss=0.1409, simple_loss=0.1663, pruned_loss=0.05769, over 2005866.13 frames. ], batch size: 99, lr: 8.64e-03, grad_scale: 8.0 2022-12-07 21:00:26,482 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:00:54,936 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 2.352e+02 2.835e+02 3.387e+02 8.466e+02, threshold=5.670e+02, percent-clipped=3.0 2022-12-07 21:01:04,676 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:01:09,005 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:01:14,314 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6349, 4.0207, 2.9446, 4.8376, 4.2945, 4.6404, 4.1650, 3.4135], device='cuda:1'), covar=tensor([0.0564, 0.1198, 0.4684, 0.0558, 0.1280, 0.1474, 0.1036, 0.3647], device='cuda:1'), in_proj_covar=tensor([0.0261, 0.0303, 0.0281, 0.0237, 0.0302, 0.0291, 0.0259, 0.0266], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 21:01:40,464 INFO [train.py:873] (1/4) Epoch 9, batch 6500, loss[loss=0.1445, simple_loss=0.1511, pruned_loss=0.06892, over 3865.00 frames. ], tot_loss[loss=0.143, simple_loss=0.1674, pruned_loss=0.05925, over 1984101.10 frames. ], batch size: 100, lr: 8.64e-03, grad_scale: 8.0 2022-12-07 21:01:48,716 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3953, 4.3705, 4.7335, 4.0844, 4.5194, 4.7738, 1.7071, 4.2445], device='cuda:1'), covar=tensor([0.0256, 0.0303, 0.0371, 0.0378, 0.0276, 0.0150, 0.2980, 0.0262], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0157, 0.0130, 0.0129, 0.0186, 0.0127, 0.0153, 0.0175], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:01:59,307 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:02:24,261 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.174e+02 2.311e+02 2.795e+02 3.453e+02 6.517e+02, threshold=5.590e+02, percent-clipped=2.0 2022-12-07 21:02:57,844 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:03:01,139 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:03:01,465 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=10.70 vs. limit=5.0 2022-12-07 21:03:10,246 INFO [train.py:873] (1/4) Epoch 9, batch 6600, loss[loss=0.1581, simple_loss=0.1788, pruned_loss=0.06871, over 14129.00 frames. ], tot_loss[loss=0.1411, simple_loss=0.1664, pruned_loss=0.05795, over 2030813.44 frames. ], batch size: 99, lr: 8.63e-03, grad_scale: 8.0 2022-12-07 21:03:10,387 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7236, 4.7177, 4.8859, 4.2967, 4.8031, 5.2712, 1.7421, 4.5433], device='cuda:1'), covar=tensor([0.0336, 0.0377, 0.0629, 0.0461, 0.0385, 0.0145, 0.3816, 0.0330], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0159, 0.0131, 0.0130, 0.0188, 0.0129, 0.0153, 0.0176], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:03:21,962 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2022-12-07 21:03:53,414 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:03:54,921 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.185e+02 2.883e+02 3.643e+02 7.006e+02, threshold=5.766e+02, percent-clipped=2.0 2022-12-07 21:04:22,324 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:04:40,786 INFO [train.py:873] (1/4) Epoch 9, batch 6700, loss[loss=0.1648, simple_loss=0.1846, pruned_loss=0.07252, over 14261.00 frames. ], tot_loss[loss=0.1421, simple_loss=0.1669, pruned_loss=0.05868, over 1993905.78 frames. ], batch size: 80, lr: 8.62e-03, grad_scale: 8.0 2022-12-07 21:04:47,626 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-07 21:05:06,522 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:05:19,215 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 21:05:24,548 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.308e+02 2.851e+02 3.420e+02 6.457e+02, threshold=5.703e+02, percent-clipped=2.0 2022-12-07 21:06:10,530 INFO [train.py:873] (1/4) Epoch 9, batch 6800, loss[loss=0.1459, simple_loss=0.1741, pruned_loss=0.05885, over 14526.00 frames. ], tot_loss[loss=0.1419, simple_loss=0.1664, pruned_loss=0.05868, over 1940152.23 frames. ], batch size: 51, lr: 8.62e-03, grad_scale: 8.0 2022-12-07 21:06:18,017 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5708, 1.4937, 1.5189, 1.6575, 1.6276, 1.1480, 1.1325, 1.0161], device='cuda:1'), covar=tensor([0.0500, 0.0442, 0.0366, 0.0256, 0.0288, 0.0278, 0.0255, 0.0526], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:1'), out_proj_covar=tensor([9.7614e-05, 1.0711e-04, 9.3423e-05, 1.0141e-04, 9.8786e-05, 1.4765e-04, 1.2471e-04, 1.4411e-04], device='cuda:1') 2022-12-07 21:06:25,261 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:06:42,808 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3261, 5.0737, 4.9072, 5.4481, 4.9806, 4.7448, 5.4447, 5.3155], device='cuda:1'), covar=tensor([0.0650, 0.0597, 0.0804, 0.0395, 0.0667, 0.0390, 0.0571, 0.0486], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0119, 0.0129, 0.0136, 0.0132, 0.0107, 0.0150, 0.0130], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 21:06:54,534 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.355e+02 2.871e+02 3.825e+02 6.310e+02, threshold=5.742e+02, percent-clipped=3.0 2022-12-07 21:07:25,515 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3158, 2.1273, 2.2830, 1.5036, 2.1056, 2.2277, 2.3912, 2.0393], device='cuda:1'), covar=tensor([0.0769, 0.0901, 0.0967, 0.1669, 0.1286, 0.0847, 0.0546, 0.1474], device='cuda:1'), in_proj_covar=tensor([0.0130, 0.0184, 0.0133, 0.0125, 0.0126, 0.0136, 0.0112, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:07:31,834 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:07:40,406 INFO [train.py:873] (1/4) Epoch 9, batch 6900, loss[loss=0.1246, simple_loss=0.1613, pruned_loss=0.04395, over 14285.00 frames. ], tot_loss[loss=0.1421, simple_loss=0.1667, pruned_loss=0.05872, over 1949491.81 frames. ], batch size: 31, lr: 8.61e-03, grad_scale: 8.0 2022-12-07 21:07:44,071 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 21:08:14,505 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:08:17,956 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:08:24,082 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.180e+02 2.447e+02 3.065e+02 3.875e+02 6.583e+02, threshold=6.130e+02, percent-clipped=2.0 2022-12-07 21:09:09,668 INFO [train.py:873] (1/4) Epoch 9, batch 7000, loss[loss=0.13, simple_loss=0.1661, pruned_loss=0.04697, over 14197.00 frames. ], tot_loss[loss=0.1417, simple_loss=0.1665, pruned_loss=0.05845, over 1941915.29 frames. ], batch size: 80, lr: 8.60e-03, grad_scale: 8.0 2022-12-07 21:09:17,394 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8033, 1.4077, 3.7524, 1.9935, 3.6967, 3.8501, 2.7764, 4.1988], device='cuda:1'), covar=tensor([0.0213, 0.3048, 0.0359, 0.1865, 0.0511, 0.0409, 0.0766, 0.0143], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0157, 0.0158, 0.0169, 0.0171, 0.0169, 0.0136, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:09:19,106 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0331, 1.4469, 3.2002, 1.5964, 3.1547, 3.1457, 2.1867, 3.3812], device='cuda:1'), covar=tensor([0.0246, 0.2722, 0.0337, 0.1935, 0.0727, 0.0440, 0.0942, 0.0204], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0157, 0.0158, 0.0169, 0.0171, 0.0169, 0.0136, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:09:24,651 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:09:38,477 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9015, 1.4825, 2.5278, 2.3761, 2.4548, 2.5252, 1.8601, 2.5040], device='cuda:1'), covar=tensor([0.0644, 0.0917, 0.0141, 0.0307, 0.0316, 0.0132, 0.0457, 0.0192], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0122, 0.0165, 0.0140, 0.0136, 0.0115, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:09:53,898 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.336e+02 2.835e+02 3.380e+02 6.445e+02, threshold=5.670e+02, percent-clipped=1.0 2022-12-07 21:10:19,392 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:10:36,239 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 21:10:40,044 INFO [train.py:873] (1/4) Epoch 9, batch 7100, loss[loss=0.1449, simple_loss=0.1623, pruned_loss=0.06371, over 6000.00 frames. ], tot_loss[loss=0.1409, simple_loss=0.1661, pruned_loss=0.05783, over 1927588.32 frames. ], batch size: 100, lr: 8.60e-03, grad_scale: 8.0 2022-12-07 21:10:55,499 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:11:27,145 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.435e+02 3.463e+02 4.295e+02 1.529e+03, threshold=6.926e+02, percent-clipped=8.0 2022-12-07 21:11:38,377 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 21:11:42,147 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:00,549 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:11,923 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:16,513 INFO [train.py:873] (1/4) Epoch 9, batch 7200, loss[loss=0.1577, simple_loss=0.1666, pruned_loss=0.07446, over 6941.00 frames. ], tot_loss[loss=0.1421, simple_loss=0.1669, pruned_loss=0.05867, over 1969329.85 frames. ], batch size: 100, lr: 8.59e-03, grad_scale: 8.0 2022-12-07 21:12:57,742 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:59,602 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:13:04,123 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.252e+02 2.773e+02 3.514e+02 7.822e+02, threshold=5.546e+02, percent-clipped=1.0 2022-12-07 21:13:11,127 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:13:43,175 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:13:52,642 INFO [train.py:873] (1/4) Epoch 9, batch 7300, loss[loss=0.1346, simple_loss=0.1705, pruned_loss=0.04934, over 14355.00 frames. ], tot_loss[loss=0.1402, simple_loss=0.1657, pruned_loss=0.05733, over 2004846.76 frames. ], batch size: 55, lr: 8.58e-03, grad_scale: 8.0 2022-12-07 21:14:03,352 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.44 vs. limit=2.0 2022-12-07 21:14:16,668 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8249, 2.4367, 3.6398, 2.6566, 3.6029, 3.4756, 3.4417, 2.7922], device='cuda:1'), covar=tensor([0.0758, 0.3358, 0.0957, 0.2039, 0.0917, 0.0814, 0.1479, 0.2180], device='cuda:1'), in_proj_covar=tensor([0.0334, 0.0318, 0.0396, 0.0300, 0.0376, 0.0313, 0.0360, 0.0317], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:14:39,307 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.094e+02 2.117e+02 2.726e+02 3.356e+02 6.858e+02, threshold=5.453e+02, percent-clipped=1.0 2022-12-07 21:15:01,842 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:15:04,086 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8229, 2.0923, 2.1204, 2.1586, 1.9607, 2.1955, 1.8737, 1.3244], device='cuda:1'), covar=tensor([0.1606, 0.0898, 0.0886, 0.0552, 0.1027, 0.0592, 0.1725, 0.2387], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0069, 0.0056, 0.0059, 0.0084, 0.0066, 0.0090, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:15:28,020 INFO [train.py:873] (1/4) Epoch 9, batch 7400, loss[loss=0.1323, simple_loss=0.1484, pruned_loss=0.05812, over 4986.00 frames. ], tot_loss[loss=0.1414, simple_loss=0.1663, pruned_loss=0.05823, over 1941117.45 frames. ], batch size: 100, lr: 8.58e-03, grad_scale: 8.0 2022-12-07 21:15:28,132 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5951, 1.3366, 2.0296, 1.8792, 1.9534, 2.0448, 1.5295, 2.0599], device='cuda:1'), covar=tensor([0.0596, 0.0869, 0.0178, 0.0344, 0.0426, 0.0170, 0.0453, 0.0225], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0123, 0.0166, 0.0142, 0.0136, 0.0114, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:16:15,149 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.396e+02 2.952e+02 3.564e+02 5.471e+02, threshold=5.904e+02, percent-clipped=1.0 2022-12-07 21:16:47,234 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:02,497 INFO [train.py:873] (1/4) Epoch 9, batch 7500, loss[loss=0.1701, simple_loss=0.181, pruned_loss=0.07961, over 10383.00 frames. ], tot_loss[loss=0.1398, simple_loss=0.1652, pruned_loss=0.05718, over 1974671.53 frames. ], batch size: 100, lr: 8.57e-03, grad_scale: 8.0 2022-12-07 21:17:38,193 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:42,606 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:44,527 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.588e+01 2.266e+02 2.777e+02 3.907e+02 7.467e+02, threshold=5.554e+02, percent-clipped=5.0 2022-12-07 21:17:45,637 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:17:45,687 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4009, 1.0419, 1.3398, 0.9170, 1.0825, 1.3653, 1.1146, 1.1983], device='cuda:1'), covar=tensor([0.0396, 0.0660, 0.0650, 0.0454, 0.1110, 0.0686, 0.0406, 0.1159], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0184, 0.0133, 0.0126, 0.0128, 0.0137, 0.0114, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:18:40,270 INFO [train.py:873] (1/4) Epoch 10, batch 0, loss[loss=0.1614, simple_loss=0.183, pruned_loss=0.06987, over 14084.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.183, pruned_loss=0.06987, over 14084.00 frames. ], batch size: 29, lr: 8.15e-03, grad_scale: 8.0 2022-12-07 21:18:40,270 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 21:18:48,289 INFO [train.py:905] (1/4) Epoch 10, validation: loss=0.1297, simple_loss=0.1728, pruned_loss=0.04327, over 857387.00 frames. 2022-12-07 21:18:48,290 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 21:19:16,402 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:20:03,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5237, 3.8051, 3.0242, 4.8353, 4.2552, 4.4737, 4.1053, 3.2740], device='cuda:1'), covar=tensor([0.0852, 0.1430, 0.4098, 0.0489, 0.0974, 0.1736, 0.1141, 0.3822], device='cuda:1'), in_proj_covar=tensor([0.0262, 0.0300, 0.0274, 0.0234, 0.0298, 0.0291, 0.0258, 0.0261], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 21:20:03,858 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8569, 1.5382, 3.7816, 3.5421, 3.5776, 3.8241, 3.1365, 3.8150], device='cuda:1'), covar=tensor([0.1291, 0.1471, 0.0102, 0.0209, 0.0204, 0.0102, 0.0240, 0.0131], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0122, 0.0166, 0.0141, 0.0135, 0.0115, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:20:12,822 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.622e+01 2.492e+02 3.120e+02 3.862e+02 1.100e+03, threshold=6.240e+02, percent-clipped=9.0 2022-12-07 21:20:14,402 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:20:24,791 INFO [train.py:873] (1/4) Epoch 10, batch 100, loss[loss=0.1509, simple_loss=0.173, pruned_loss=0.06441, over 13528.00 frames. ], tot_loss[loss=0.1412, simple_loss=0.1676, pruned_loss=0.05746, over 897606.52 frames. ], batch size: 100, lr: 8.14e-03, grad_scale: 4.0 2022-12-07 21:20:34,432 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:20:38,225 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7030, 2.3373, 3.2157, 2.2041, 2.0376, 2.6768, 1.2812, 2.7221], device='cuda:1'), covar=tensor([0.1180, 0.1644, 0.0927, 0.2838, 0.2996, 0.1310, 0.5295, 0.1166], device='cuda:1'), in_proj_covar=tensor([0.0076, 0.0091, 0.0086, 0.0092, 0.0111, 0.0079, 0.0126, 0.0084], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:21:19,966 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:21:48,939 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.529e+02 3.051e+02 3.928e+02 8.118e+02, threshold=6.102e+02, percent-clipped=5.0 2022-12-07 21:21:59,798 INFO [train.py:873] (1/4) Epoch 10, batch 200, loss[loss=0.1178, simple_loss=0.1496, pruned_loss=0.04298, over 13957.00 frames. ], tot_loss[loss=0.1388, simple_loss=0.1658, pruned_loss=0.05593, over 1378846.66 frames. ], batch size: 19, lr: 8.14e-03, grad_scale: 4.0 2022-12-07 21:22:17,989 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4537, 3.1794, 3.1689, 3.4419, 3.3061, 3.4367, 3.5234, 2.9296], device='cuda:1'), covar=tensor([0.0469, 0.1126, 0.0526, 0.0601, 0.0775, 0.0397, 0.0632, 0.0672], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0248, 0.0173, 0.0168, 0.0166, 0.0134, 0.0253, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 21:22:26,725 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9561, 1.5844, 4.0773, 3.7630, 3.8003, 4.0718, 3.4455, 4.1012], device='cuda:1'), covar=tensor([0.1307, 0.1382, 0.0091, 0.0215, 0.0195, 0.0107, 0.0225, 0.0125], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0122, 0.0167, 0.0142, 0.0136, 0.0115, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:23:12,742 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:13,638 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:23,109 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 2.072e+02 2.682e+02 3.537e+02 7.107e+02, threshold=5.365e+02, percent-clipped=2.0 2022-12-07 21:23:24,317 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:34,540 INFO [train.py:873] (1/4) Epoch 10, batch 300, loss[loss=0.1261, simple_loss=0.1606, pruned_loss=0.04577, over 14227.00 frames. ], tot_loss[loss=0.139, simple_loss=0.1653, pruned_loss=0.05638, over 1608126.27 frames. ], batch size: 60, lr: 8.13e-03, grad_scale: 4.0 2022-12-07 21:23:57,993 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:10,432 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:11,139 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 21:24:13,362 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:31,213 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2022-12-07 21:24:56,059 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:24:59,359 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.338e+02 2.858e+02 3.586e+02 7.542e+02, threshold=5.716e+02, percent-clipped=9.0 2022-12-07 21:25:08,420 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7864, 3.1451, 4.4509, 3.3303, 4.4171, 4.4487, 4.1882, 3.8095], device='cuda:1'), covar=tensor([0.0605, 0.2976, 0.1106, 0.1795, 0.0959, 0.0707, 0.1636, 0.1995], device='cuda:1'), in_proj_covar=tensor([0.0336, 0.0319, 0.0401, 0.0304, 0.0372, 0.0314, 0.0363, 0.0320], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:25:10,941 INFO [train.py:873] (1/4) Epoch 10, batch 400, loss[loss=0.1304, simple_loss=0.1678, pruned_loss=0.0465, over 14250.00 frames. ], tot_loss[loss=0.1383, simple_loss=0.1646, pruned_loss=0.05598, over 1730351.49 frames. ], batch size: 37, lr: 8.12e-03, grad_scale: 8.0 2022-12-07 21:25:12,109 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:26:35,457 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.435e+02 2.815e+02 3.563e+02 7.495e+02, threshold=5.630e+02, percent-clipped=4.0 2022-12-07 21:26:47,612 INFO [train.py:873] (1/4) Epoch 10, batch 500, loss[loss=0.1438, simple_loss=0.1639, pruned_loss=0.06188, over 6928.00 frames. ], tot_loss[loss=0.137, simple_loss=0.1639, pruned_loss=0.0551, over 1819875.07 frames. ], batch size: 100, lr: 8.12e-03, grad_scale: 8.0 2022-12-07 21:26:55,051 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 21:27:05,940 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:27:18,159 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6489, 1.4511, 3.6653, 1.7304, 3.6709, 3.8261, 2.6783, 4.1069], device='cuda:1'), covar=tensor([0.0264, 0.3234, 0.0422, 0.2274, 0.0520, 0.0412, 0.0907, 0.0170], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0159, 0.0159, 0.0170, 0.0171, 0.0171, 0.0137, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:27:41,605 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8779, 3.6830, 3.3185, 2.6404, 3.3087, 3.6342, 4.0609, 3.2489], device='cuda:1'), covar=tensor([0.0570, 0.1880, 0.1132, 0.1959, 0.0839, 0.0557, 0.0708, 0.1413], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0187, 0.0133, 0.0126, 0.0127, 0.0136, 0.0113, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:28:01,939 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:03,538 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:10,691 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.029e+02 2.141e+02 2.717e+02 3.371e+02 6.429e+02, threshold=5.434e+02, percent-clipped=1.0 2022-12-07 21:28:21,923 INFO [train.py:873] (1/4) Epoch 10, batch 600, loss[loss=0.1291, simple_loss=0.1642, pruned_loss=0.04703, over 14332.00 frames. ], tot_loss[loss=0.137, simple_loss=0.1638, pruned_loss=0.05505, over 1918107.12 frames. ], batch size: 73, lr: 8.11e-03, grad_scale: 8.0 2022-12-07 21:28:46,813 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:46,933 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:53,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5954, 3.4376, 3.2000, 2.4298, 3.0011, 3.3961, 3.5895, 2.8759], device='cuda:1'), covar=tensor([0.0529, 0.1450, 0.1014, 0.1712, 0.0910, 0.0561, 0.0775, 0.1404], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0188, 0.0135, 0.0128, 0.0128, 0.0138, 0.0113, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 21:29:41,783 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:29:44,890 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:29:45,571 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.288e+02 3.002e+02 3.760e+02 7.437e+02, threshold=6.005e+02, percent-clipped=6.0 2022-12-07 21:29:53,318 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:29:53,833 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-07 21:29:57,180 INFO [train.py:873] (1/4) Epoch 10, batch 700, loss[loss=0.1765, simple_loss=0.1832, pruned_loss=0.08494, over 8620.00 frames. ], tot_loss[loss=0.1376, simple_loss=0.164, pruned_loss=0.0556, over 1937141.90 frames. ], batch size: 100, lr: 8.11e-03, grad_scale: 8.0 2022-12-07 21:30:03,683 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1758, 3.0782, 2.7788, 2.8030, 3.2347, 3.1454, 3.3081, 3.2132], device='cuda:1'), covar=tensor([0.1467, 0.0880, 0.2644, 0.3427, 0.1102, 0.1391, 0.1409, 0.1362], device='cuda:1'), in_proj_covar=tensor([0.0357, 0.0251, 0.0423, 0.0535, 0.0312, 0.0407, 0.0392, 0.0354], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:30:27,493 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:30:50,840 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 21:31:20,659 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-07 21:31:20,813 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.251e+02 2.713e+02 3.475e+02 5.744e+02, threshold=5.426e+02, percent-clipped=0.0 2022-12-07 21:31:32,392 INFO [train.py:873] (1/4) Epoch 10, batch 800, loss[loss=0.1396, simple_loss=0.174, pruned_loss=0.05259, over 14229.00 frames. ], tot_loss[loss=0.138, simple_loss=0.1641, pruned_loss=0.05597, over 1963340.20 frames. ], batch size: 35, lr: 8.10e-03, grad_scale: 8.0 2022-12-07 21:31:56,561 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 21:32:43,307 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:32:55,163 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.204e+02 2.859e+02 3.964e+02 7.134e+02, threshold=5.717e+02, percent-clipped=4.0 2022-12-07 21:33:01,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 21:33:06,466 INFO [train.py:873] (1/4) Epoch 10, batch 900, loss[loss=0.1083, simple_loss=0.1472, pruned_loss=0.03467, over 13975.00 frames. ], tot_loss[loss=0.139, simple_loss=0.1648, pruned_loss=0.05657, over 1973029.28 frames. ], batch size: 26, lr: 8.09e-03, grad_scale: 8.0 2022-12-07 21:33:44,535 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6738, 1.8038, 1.2958, 1.8250, 1.7060, 1.3272, 1.4452, 1.5878], device='cuda:1'), covar=tensor([0.0444, 0.0599, 0.0740, 0.0456, 0.0493, 0.0661, 0.0533, 0.1013], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0015, 0.0013, 0.0013, 0.0013, 0.0021, 0.0018, 0.0022], device='cuda:1'), out_proj_covar=tensor([1.0188e-04, 1.1026e-04, 9.7707e-05, 1.0470e-04, 1.0245e-04, 1.5256e-04, 1.2974e-04, 1.4704e-04], device='cuda:1') 2022-12-07 21:33:51,780 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2022-12-07 21:34:24,757 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:26,763 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:30,091 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 2.392e+02 2.888e+02 3.362e+02 7.513e+02, threshold=5.776e+02, percent-clipped=4.0 2022-12-07 21:34:37,773 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:41,174 INFO [train.py:873] (1/4) Epoch 10, batch 1000, loss[loss=0.1361, simple_loss=0.1644, pruned_loss=0.05392, over 14390.00 frames. ], tot_loss[loss=0.1386, simple_loss=0.1648, pruned_loss=0.05622, over 2019991.57 frames. ], batch size: 53, lr: 8.09e-03, grad_scale: 8.0 2022-12-07 21:35:04,095 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-07 21:35:04,672 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0890, 2.1594, 4.7380, 4.3963, 4.2858, 4.8392, 4.5465, 4.9128], device='cuda:1'), covar=tensor([0.1285, 0.1198, 0.0083, 0.0151, 0.0158, 0.0084, 0.0099, 0.0086], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0157, 0.0122, 0.0165, 0.0142, 0.0136, 0.0116, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:35:08,592 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3869, 3.5468, 5.0840, 4.0461, 5.0475, 5.2760, 4.9714, 4.7167], device='cuda:1'), covar=tensor([0.0437, 0.2158, 0.0600, 0.1199, 0.0539, 0.0443, 0.1487, 0.1482], device='cuda:1'), in_proj_covar=tensor([0.0335, 0.0318, 0.0393, 0.0302, 0.0369, 0.0311, 0.0357, 0.0315], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:35:18,451 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 21:35:22,728 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:35:23,634 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:36:04,003 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.114e+02 2.200e+02 2.692e+02 3.189e+02 5.615e+02, threshold=5.384e+02, percent-clipped=0.0 2022-12-07 21:36:13,217 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 21:36:15,496 INFO [train.py:873] (1/4) Epoch 10, batch 1100, loss[loss=0.1286, simple_loss=0.1668, pruned_loss=0.04517, over 14492.00 frames. ], tot_loss[loss=0.1391, simple_loss=0.165, pruned_loss=0.05659, over 2033142.11 frames. ], batch size: 34, lr: 8.08e-03, grad_scale: 8.0 2022-12-07 21:36:18,821 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:36:23,290 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7421, 4.6979, 5.1499, 4.1294, 4.8229, 5.3287, 2.2081, 4.4450], device='cuda:1'), covar=tensor([0.0234, 0.0339, 0.0344, 0.0488, 0.0290, 0.0090, 0.2845, 0.0329], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0160, 0.0131, 0.0132, 0.0189, 0.0130, 0.0154, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:36:24,355 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:17,002 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:18,949 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:22,768 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:27,482 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:37,918 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:37:39,433 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.182e+02 2.689e+02 3.231e+02 7.876e+02, threshold=5.378e+02, percent-clipped=5.0 2022-12-07 21:37:51,123 INFO [train.py:873] (1/4) Epoch 10, batch 1200, loss[loss=0.1557, simple_loss=0.1702, pruned_loss=0.07059, over 7799.00 frames. ], tot_loss[loss=0.1383, simple_loss=0.1648, pruned_loss=0.05587, over 2073175.42 frames. ], batch size: 100, lr: 8.08e-03, grad_scale: 8.0 2022-12-07 21:37:52,223 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2180, 3.4095, 4.2135, 2.8662, 2.3829, 3.3132, 2.0929, 3.5030], device='cuda:1'), covar=tensor([0.1481, 0.0943, 0.0498, 0.2110, 0.2609, 0.1095, 0.3486, 0.0907], device='cuda:1'), in_proj_covar=tensor([0.0078, 0.0094, 0.0087, 0.0094, 0.0114, 0.0079, 0.0129, 0.0085], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:1') 2022-12-07 21:38:03,960 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8292, 2.7750, 2.0269, 2.8815, 2.6086, 2.7521, 2.4756, 2.2466], device='cuda:1'), covar=tensor([0.0864, 0.1086, 0.3214, 0.0611, 0.0985, 0.0960, 0.1349, 0.2511], device='cuda:1'), in_proj_covar=tensor([0.0258, 0.0292, 0.0271, 0.0236, 0.0294, 0.0286, 0.0256, 0.0257], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 21:38:04,656 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:13,079 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:16,858 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:36,076 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:38:48,353 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1455, 1.2506, 1.3217, 0.9940, 1.0354, 0.8333, 0.9146, 0.8289], device='cuda:1'), covar=tensor([0.0284, 0.0320, 0.0201, 0.0236, 0.0250, 0.0449, 0.0309, 0.0609], device='cuda:1'), in_proj_covar=tensor([0.0013, 0.0015, 0.0013, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:1'), out_proj_covar=tensor([1.0207e-04, 1.1013e-04, 9.6989e-05, 1.0405e-04, 1.0239e-04, 1.5131e-04, 1.2861e-04, 1.4601e-04], device='cuda:1') 2022-12-07 21:38:49,284 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:02,139 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:39:08,981 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:14,523 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 2.336e+02 2.846e+02 3.621e+02 9.526e+02, threshold=5.693e+02, percent-clipped=8.0 2022-12-07 21:39:25,827 INFO [train.py:873] (1/4) Epoch 10, batch 1300, loss[loss=0.1358, simple_loss=0.1709, pruned_loss=0.0504, over 14284.00 frames. ], tot_loss[loss=0.1391, simple_loss=0.1646, pruned_loss=0.05682, over 1963603.50 frames. ], batch size: 80, lr: 8.07e-03, grad_scale: 8.0 2022-12-07 21:39:34,590 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:46,994 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:54,588 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:59,662 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:03,994 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:32,730 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:49,782 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.414e+02 2.899e+02 3.828e+02 8.841e+02, threshold=5.798e+02, percent-clipped=2.0 2022-12-07 21:40:52,668 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3934, 1.7194, 1.8856, 1.7796, 1.6918, 1.9090, 1.4985, 1.2576], device='cuda:1'), covar=tensor([0.1464, 0.1221, 0.0556, 0.0422, 0.1235, 0.0642, 0.2005, 0.2775], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0074, 0.0058, 0.0062, 0.0089, 0.0069, 0.0094, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:1') 2022-12-07 21:40:57,632 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:41:01,160 INFO [train.py:873] (1/4) Epoch 10, batch 1400, loss[loss=0.134, simple_loss=0.163, pruned_loss=0.05245, over 14227.00 frames. ], tot_loss[loss=0.1375, simple_loss=0.1639, pruned_loss=0.05552, over 1972702.54 frames. ], batch size: 94, lr: 8.07e-03, grad_scale: 8.0 2022-12-07 21:41:16,860 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9637, 1.6390, 3.1345, 2.1069, 3.0315, 1.6130, 2.3124, 2.8978], device='cuda:1'), covar=tensor([0.1199, 0.5708, 0.0664, 0.7484, 0.1003, 0.5104, 0.1777, 0.0856], device='cuda:1'), in_proj_covar=tensor([0.0243, 0.0221, 0.0193, 0.0299, 0.0218, 0.0228, 0.0219, 0.0201], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:41:49,766 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 21:41:57,164 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:42:02,611 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:42:04,189 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 21:42:15,345 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.07 vs. limit=2.0 2022-12-07 21:42:18,521 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8467, 1.4438, 3.1233, 1.3833, 3.2051, 3.0882, 2.2590, 3.1912], device='cuda:1'), covar=tensor([0.0366, 0.3086, 0.0401, 0.2482, 0.0375, 0.0451, 0.0991, 0.0276], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0159, 0.0158, 0.0169, 0.0171, 0.0171, 0.0135, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:42:24,551 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.449e+02 2.299e+02 2.933e+02 3.622e+02 8.157e+02, threshold=5.865e+02, percent-clipped=2.0 2022-12-07 21:42:35,973 INFO [train.py:873] (1/4) Epoch 10, batch 1500, loss[loss=0.1532, simple_loss=0.1506, pruned_loss=0.07786, over 2650.00 frames. ], tot_loss[loss=0.138, simple_loss=0.1641, pruned_loss=0.05599, over 1957825.73 frames. ], batch size: 100, lr: 8.06e-03, grad_scale: 8.0 2022-12-07 21:42:40,586 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-07 21:42:57,194 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:43:15,884 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:43:42,180 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:43:59,277 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.142e+02 2.757e+02 3.564e+02 6.285e+02, threshold=5.515e+02, percent-clipped=1.0 2022-12-07 21:44:10,746 INFO [train.py:873] (1/4) Epoch 10, batch 1600, loss[loss=0.1558, simple_loss=0.1686, pruned_loss=0.07148, over 13880.00 frames. ], tot_loss[loss=0.1389, simple_loss=0.1639, pruned_loss=0.0569, over 1883436.81 frames. ], batch size: 20, lr: 8.05e-03, grad_scale: 8.0 2022-12-07 21:44:24,872 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:44:26,616 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:44:48,109 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:11,273 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:20,075 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8595, 1.3265, 2.0193, 1.3621, 1.9854, 2.0673, 1.8066, 2.0993], device='cuda:1'), covar=tensor([0.0372, 0.2035, 0.0413, 0.1715, 0.0503, 0.0484, 0.0893, 0.0327], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0160, 0.0157, 0.0168, 0.0170, 0.0169, 0.0135, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:45:21,894 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:45:33,107 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.146e+01 2.291e+02 2.751e+02 3.642e+02 1.145e+03, threshold=5.503e+02, percent-clipped=2.0 2022-12-07 21:45:33,205 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:36,295 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:45,008 INFO [train.py:873] (1/4) Epoch 10, batch 1700, loss[loss=0.1229, simple_loss=0.1563, pruned_loss=0.04474, over 14131.00 frames. ], tot_loss[loss=0.1378, simple_loss=0.1634, pruned_loss=0.05607, over 1930781.29 frames. ], batch size: 99, lr: 8.05e-03, grad_scale: 8.0 2022-12-07 21:46:22,161 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9901, 2.9085, 2.0604, 3.0484, 2.8942, 2.9028, 2.5789, 2.3401], device='cuda:1'), covar=tensor([0.0935, 0.1422, 0.3478, 0.0592, 0.0886, 0.1208, 0.1411, 0.2976], device='cuda:1'), in_proj_covar=tensor([0.0264, 0.0297, 0.0271, 0.0241, 0.0297, 0.0292, 0.0259, 0.0259], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 21:46:40,947 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:46:46,191 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:07,195 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-07 21:47:08,315 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.221e+02 2.761e+02 3.390e+02 7.818e+02, threshold=5.522e+02, percent-clipped=1.0 2022-12-07 21:47:19,440 INFO [train.py:873] (1/4) Epoch 10, batch 1800, loss[loss=0.1554, simple_loss=0.151, pruned_loss=0.07995, over 2584.00 frames. ], tot_loss[loss=0.1371, simple_loss=0.163, pruned_loss=0.05561, over 1881038.43 frames. ], batch size: 100, lr: 8.04e-03, grad_scale: 8.0 2022-12-07 21:47:26,411 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:32,292 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:32,451 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3560, 1.9593, 2.3675, 2.4619, 2.2480, 1.9181, 2.4543, 2.1250], device='cuda:1'), covar=tensor([0.0225, 0.0467, 0.0260, 0.0228, 0.0279, 0.0618, 0.0232, 0.0378], device='cuda:1'), in_proj_covar=tensor([0.0270, 0.0241, 0.0361, 0.0304, 0.0248, 0.0294, 0.0276, 0.0272], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 21:47:40,981 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:59,696 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:48:15,047 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 21:48:26,764 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:48:26,862 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:48:28,692 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3977, 1.0858, 1.3012, 0.9571, 1.0260, 1.3332, 1.1186, 1.1361], device='cuda:1'), covar=tensor([0.0397, 0.0793, 0.0710, 0.0580, 0.0967, 0.0774, 0.0358, 0.1186], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0186, 0.0134, 0.0125, 0.0127, 0.0137, 0.0113, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:48:43,307 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.361e+02 2.916e+02 3.771e+02 6.970e+02, threshold=5.833e+02, percent-clipped=3.0 2022-12-07 21:48:45,619 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:48:55,139 INFO [train.py:873] (1/4) Epoch 10, batch 1900, loss[loss=0.1206, simple_loss=0.1587, pruned_loss=0.04129, over 14566.00 frames. ], tot_loss[loss=0.1374, simple_loss=0.1636, pruned_loss=0.05555, over 1935626.08 frames. ], batch size: 23, lr: 8.04e-03, grad_scale: 8.0 2022-12-07 21:49:11,035 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:49:11,847 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:02,117 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:02,186 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:08,097 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:50:24,070 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.053e+02 2.126e+02 2.650e+02 3.300e+02 6.481e+02, threshold=5.300e+02, percent-clipped=2.0 2022-12-07 21:50:27,048 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:30,522 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.35 vs. limit=5.0 2022-12-07 21:50:35,178 INFO [train.py:873] (1/4) Epoch 10, batch 2000, loss[loss=0.1629, simple_loss=0.1709, pruned_loss=0.07745, over 6002.00 frames. ], tot_loss[loss=0.1382, simple_loss=0.1642, pruned_loss=0.05608, over 1917239.16 frames. ], batch size: 100, lr: 8.03e-03, grad_scale: 8.0 2022-12-07 21:50:35,809 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6344, 3.1812, 2.5012, 3.7955, 3.7389, 3.6213, 2.9968, 2.5780], device='cuda:1'), covar=tensor([0.1028, 0.1624, 0.3881, 0.0614, 0.0757, 0.1609, 0.1582, 0.4358], device='cuda:1'), in_proj_covar=tensor([0.0261, 0.0297, 0.0270, 0.0242, 0.0297, 0.0289, 0.0259, 0.0258], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 21:50:48,010 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:12,863 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:24,473 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:29,348 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3647, 2.8718, 2.9695, 1.8913, 2.9565, 3.1590, 3.4457, 2.5547], device='cuda:1'), covar=tensor([0.0770, 0.2390, 0.1182, 0.2601, 0.1024, 0.0770, 0.0779, 0.2030], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0186, 0.0134, 0.0126, 0.0127, 0.0137, 0.0113, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:51:30,273 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3275, 2.9768, 3.0458, 2.0907, 2.8164, 2.9917, 3.2819, 2.6987], device='cuda:1'), covar=tensor([0.0678, 0.1790, 0.1111, 0.2147, 0.1151, 0.0942, 0.0777, 0.1628], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0186, 0.0134, 0.0126, 0.0127, 0.0137, 0.0113, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:51:45,851 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 21:51:59,346 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.415e+02 2.238e+02 2.810e+02 3.599e+02 8.591e+02, threshold=5.620e+02, percent-clipped=6.0 2022-12-07 21:52:10,762 INFO [train.py:873] (1/4) Epoch 10, batch 2100, loss[loss=0.1365, simple_loss=0.1621, pruned_loss=0.05545, over 14205.00 frames. ], tot_loss[loss=0.1387, simple_loss=0.1648, pruned_loss=0.05628, over 1970372.24 frames. ], batch size: 94, lr: 8.03e-03, grad_scale: 16.0 2022-12-07 21:52:17,596 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8986, 2.9959, 3.8619, 2.7578, 2.4607, 3.4382, 1.7860, 3.5147], device='cuda:1'), covar=tensor([0.1334, 0.0982, 0.0552, 0.1837, 0.2251, 0.0610, 0.4387, 0.0613], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0093, 0.0087, 0.0094, 0.0112, 0.0079, 0.0127, 0.0083], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:1') 2022-12-07 21:52:22,362 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:52:39,562 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-07 21:53:20,326 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3637, 4.2072, 3.8236, 3.9863, 4.2381, 4.3166, 4.3951, 4.3741], device='cuda:1'), covar=tensor([0.0931, 0.0509, 0.2143, 0.2718, 0.0719, 0.0763, 0.0920, 0.0896], device='cuda:1'), in_proj_covar=tensor([0.0359, 0.0246, 0.0415, 0.0527, 0.0312, 0.0408, 0.0387, 0.0351], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:53:34,947 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.579e+01 2.218e+02 2.863e+02 3.700e+02 6.629e+02, threshold=5.727e+02, percent-clipped=4.0 2022-12-07 21:53:45,271 INFO [train.py:873] (1/4) Epoch 10, batch 2200, loss[loss=0.1607, simple_loss=0.1592, pruned_loss=0.08109, over 3898.00 frames. ], tot_loss[loss=0.1389, simple_loss=0.1651, pruned_loss=0.05635, over 2022135.60 frames. ], batch size: 100, lr: 8.02e-03, grad_scale: 8.0 2022-12-07 21:53:46,782 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:54:20,783 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3297, 2.5916, 2.6478, 2.7088, 2.0897, 2.6958, 2.4303, 1.1921], device='cuda:1'), covar=tensor([0.1957, 0.0997, 0.0736, 0.0538, 0.1258, 0.0737, 0.1347, 0.3126], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0075, 0.0060, 0.0062, 0.0090, 0.0070, 0.0095, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0006, 0.0006], device='cuda:1') 2022-12-07 21:54:43,891 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:54:52,766 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:55:09,312 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.038e+02 2.253e+02 2.794e+02 3.378e+02 9.827e+02, threshold=5.588e+02, percent-clipped=3.0 2022-12-07 21:55:19,408 INFO [train.py:873] (1/4) Epoch 10, batch 2300, loss[loss=0.1812, simple_loss=0.1643, pruned_loss=0.09911, over 1247.00 frames. ], tot_loss[loss=0.1366, simple_loss=0.1633, pruned_loss=0.05489, over 2053891.75 frames. ], batch size: 100, lr: 8.01e-03, grad_scale: 8.0 2022-12-07 21:55:37,466 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:56:39,698 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0781, 2.6630, 3.8111, 2.8707, 3.9305, 3.7137, 3.7031, 3.2671], device='cuda:1'), covar=tensor([0.0766, 0.2721, 0.0973, 0.2006, 0.0862, 0.0860, 0.1470, 0.1983], device='cuda:1'), in_proj_covar=tensor([0.0336, 0.0316, 0.0391, 0.0304, 0.0373, 0.0312, 0.0356, 0.0313], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 21:56:43,385 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.243e+01 2.133e+02 2.592e+02 3.520e+02 8.171e+02, threshold=5.183e+02, percent-clipped=4.0 2022-12-07 21:56:53,786 INFO [train.py:873] (1/4) Epoch 10, batch 2400, loss[loss=0.1268, simple_loss=0.1313, pruned_loss=0.06119, over 2647.00 frames. ], tot_loss[loss=0.1378, simple_loss=0.1642, pruned_loss=0.05565, over 2037204.67 frames. ], batch size: 100, lr: 8.01e-03, grad_scale: 8.0 2022-12-07 21:57:00,712 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:57:32,410 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-07 21:57:37,090 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8765, 1.6102, 1.9566, 1.3431, 1.7190, 1.9309, 1.8765, 1.6408], device='cuda:1'), covar=tensor([0.0879, 0.0866, 0.0868, 0.1320, 0.1322, 0.0669, 0.0570, 0.2145], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0185, 0.0135, 0.0125, 0.0126, 0.0136, 0.0112, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 21:58:14,472 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0424, 0.8925, 0.8802, 0.9041, 1.0723, 0.5699, 0.8602, 0.9902], device='cuda:1'), covar=tensor([0.0445, 0.0749, 0.0428, 0.0536, 0.0341, 0.0605, 0.0596, 0.0599], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0023, 0.0026, 0.0023, 0.0025, 0.0035, 0.0024, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 21:58:18,046 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.395e+02 3.014e+02 3.583e+02 8.568e+02, threshold=6.028e+02, percent-clipped=4.0 2022-12-07 21:58:28,703 INFO [train.py:873] (1/4) Epoch 10, batch 2500, loss[loss=0.1229, simple_loss=0.1554, pruned_loss=0.04518, over 13542.00 frames. ], tot_loss[loss=0.1362, simple_loss=0.1633, pruned_loss=0.05455, over 2029251.54 frames. ], batch size: 100, lr: 8.00e-03, grad_scale: 8.0 2022-12-07 21:59:13,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1772, 1.3094, 1.4108, 0.9523, 0.8515, 1.2063, 0.8661, 1.3103], device='cuda:1'), covar=tensor([0.1233, 0.2900, 0.0828, 0.2426, 0.3197, 0.0953, 0.2813, 0.0930], device='cuda:1'), in_proj_covar=tensor([0.0077, 0.0092, 0.0085, 0.0092, 0.0110, 0.0078, 0.0126, 0.0083], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:59:22,845 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:59:33,231 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3037, 1.3273, 3.4388, 1.4102, 3.2198, 3.4572, 2.4073, 3.6802], device='cuda:1'), covar=tensor([0.0250, 0.3138, 0.0345, 0.2490, 0.0825, 0.0399, 0.0895, 0.0172], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0158, 0.0156, 0.0168, 0.0168, 0.0168, 0.0134, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 21:59:42,636 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2022-12-07 21:59:49,681 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.39 vs. limit=5.0 2022-12-07 21:59:53,710 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.220e+02 2.772e+02 3.507e+02 5.637e+02, threshold=5.543e+02, percent-clipped=0.0 2022-12-07 22:00:03,810 INFO [train.py:873] (1/4) Epoch 10, batch 2600, loss[loss=0.168, simple_loss=0.187, pruned_loss=0.07456, over 13549.00 frames. ], tot_loss[loss=0.1364, simple_loss=0.1632, pruned_loss=0.05474, over 1967272.00 frames. ], batch size: 100, lr: 8.00e-03, grad_scale: 8.0 2022-12-07 22:00:14,564 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1781, 2.9505, 2.8386, 1.8309, 2.6676, 2.9019, 3.3361, 2.5427], device='cuda:1'), covar=tensor([0.0726, 0.1391, 0.1216, 0.2063, 0.1146, 0.0801, 0.0572, 0.1623], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0184, 0.0134, 0.0126, 0.0126, 0.0137, 0.0112, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 22:01:27,326 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.257e+02 2.909e+02 3.862e+02 1.406e+03, threshold=5.817e+02, percent-clipped=9.0 2022-12-07 22:01:38,026 INFO [train.py:873] (1/4) Epoch 10, batch 2700, loss[loss=0.1741, simple_loss=0.1827, pruned_loss=0.0828, over 10318.00 frames. ], tot_loss[loss=0.137, simple_loss=0.1639, pruned_loss=0.05509, over 1992538.97 frames. ], batch size: 100, lr: 7.99e-03, grad_scale: 8.0 2022-12-07 22:01:45,096 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:01:49,247 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 22:02:30,606 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:02:41,527 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 22:03:03,174 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.053e+02 2.251e+02 2.633e+02 3.197e+02 4.846e+02, threshold=5.267e+02, percent-clipped=0.0 2022-12-07 22:03:13,060 INFO [train.py:873] (1/4) Epoch 10, batch 2800, loss[loss=0.2216, simple_loss=0.1878, pruned_loss=0.1277, over 1205.00 frames. ], tot_loss[loss=0.1379, simple_loss=0.1642, pruned_loss=0.05579, over 1913112.04 frames. ], batch size: 100, lr: 7.99e-03, grad_scale: 8.0 2022-12-07 22:04:08,191 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:04:38,139 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.119e+02 2.647e+02 3.407e+02 7.312e+02, threshold=5.294e+02, percent-clipped=4.0 2022-12-07 22:04:48,430 INFO [train.py:873] (1/4) Epoch 10, batch 2900, loss[loss=0.1216, simple_loss=0.1539, pruned_loss=0.04471, over 14480.00 frames. ], tot_loss[loss=0.1367, simple_loss=0.1636, pruned_loss=0.05487, over 2046661.45 frames. ], batch size: 51, lr: 7.98e-03, grad_scale: 8.0 2022-12-07 22:04:52,931 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:04:58,466 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2022-12-07 22:05:19,657 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:05:47,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5470, 2.9156, 2.6722, 2.6845, 2.2062, 2.9327, 2.5643, 1.3196], device='cuda:1'), covar=tensor([0.1914, 0.0687, 0.1246, 0.0872, 0.1242, 0.0656, 0.1364, 0.3198], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0074, 0.0060, 0.0063, 0.0089, 0.0070, 0.0094, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:1') 2022-12-07 22:05:53,206 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 22:05:55,142 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:06:03,120 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2022-12-07 22:06:12,376 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.365e+02 2.871e+02 3.545e+02 4.870e+02, threshold=5.742e+02, percent-clipped=0.0 2022-12-07 22:06:17,184 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:06:19,501 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 22:06:22,388 INFO [train.py:873] (1/4) Epoch 10, batch 3000, loss[loss=0.1239, simple_loss=0.158, pruned_loss=0.04489, over 13994.00 frames. ], tot_loss[loss=0.1371, simple_loss=0.1636, pruned_loss=0.05532, over 2016235.14 frames. ], batch size: 22, lr: 7.98e-03, grad_scale: 8.0 2022-12-07 22:06:22,389 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 22:06:39,846 INFO [train.py:905] (1/4) Epoch 10, validation: loss=0.1251, simple_loss=0.1669, pruned_loss=0.04162, over 857387.00 frames. 2022-12-07 22:06:39,846 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 22:07:09,973 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:07:34,936 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3538, 2.1283, 2.2994, 1.5106, 2.0428, 2.3281, 2.4114, 2.0222], device='cuda:1'), covar=tensor([0.0768, 0.0740, 0.0876, 0.1731, 0.1113, 0.0620, 0.0464, 0.1409], device='cuda:1'), in_proj_covar=tensor([0.0130, 0.0181, 0.0131, 0.0125, 0.0126, 0.0136, 0.0112, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 22:07:41,386 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2049, 2.9481, 2.9163, 2.0479, 2.6838, 2.9088, 3.2396, 2.6082], device='cuda:1'), covar=tensor([0.0678, 0.1076, 0.0859, 0.1824, 0.0961, 0.0808, 0.0818, 0.1283], device='cuda:1'), in_proj_covar=tensor([0.0130, 0.0181, 0.0131, 0.0125, 0.0126, 0.0136, 0.0112, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 22:08:04,099 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.301e+01 2.256e+02 2.878e+02 3.587e+02 6.087e+02, threshold=5.757e+02, percent-clipped=3.0 2022-12-07 22:08:14,730 INFO [train.py:873] (1/4) Epoch 10, batch 3100, loss[loss=0.1793, simple_loss=0.1663, pruned_loss=0.09613, over 2643.00 frames. ], tot_loss[loss=0.1366, simple_loss=0.1633, pruned_loss=0.05491, over 1993228.77 frames. ], batch size: 100, lr: 7.97e-03, grad_scale: 8.0 2022-12-07 22:08:30,136 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3334, 2.2237, 4.3745, 2.9276, 4.1102, 2.1197, 3.2540, 4.2041], device='cuda:1'), covar=tensor([0.0513, 0.4476, 0.0356, 0.6691, 0.0554, 0.3572, 0.1291, 0.0346], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0223, 0.0196, 0.0299, 0.0215, 0.0225, 0.0220, 0.0204], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:08:38,181 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 22:09:38,774 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.264e+02 3.028e+02 3.832e+02 6.668e+02, threshold=6.056e+02, percent-clipped=4.0 2022-12-07 22:09:49,302 INFO [train.py:873] (1/4) Epoch 10, batch 3200, loss[loss=0.1422, simple_loss=0.1759, pruned_loss=0.05426, over 14507.00 frames. ], tot_loss[loss=0.1361, simple_loss=0.1632, pruned_loss=0.0545, over 1979434.33 frames. ], batch size: 49, lr: 7.96e-03, grad_scale: 8.0 2022-12-07 22:10:31,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1907, 1.9843, 4.9745, 4.4864, 4.4090, 5.1821, 4.9143, 5.1696], device='cuda:1'), covar=tensor([0.1226, 0.1273, 0.0078, 0.0137, 0.0172, 0.0079, 0.0080, 0.0085], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0122, 0.0165, 0.0142, 0.0132, 0.0114, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:11:13,456 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.362e+02 2.792e+02 3.327e+02 7.945e+02, threshold=5.585e+02, percent-clipped=1.0 2022-12-07 22:11:13,600 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:11:24,236 INFO [train.py:873] (1/4) Epoch 10, batch 3300, loss[loss=0.1629, simple_loss=0.1554, pruned_loss=0.0852, over 1268.00 frames. ], tot_loss[loss=0.1368, simple_loss=0.1634, pruned_loss=0.05506, over 1930905.97 frames. ], batch size: 100, lr: 7.96e-03, grad_scale: 8.0 2022-12-07 22:11:49,870 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:12:49,402 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.167e+02 2.744e+02 3.343e+02 6.323e+02, threshold=5.487e+02, percent-clipped=2.0 2022-12-07 22:12:59,544 INFO [train.py:873] (1/4) Epoch 10, batch 3400, loss[loss=0.1201, simple_loss=0.1454, pruned_loss=0.04736, over 6937.00 frames. ], tot_loss[loss=0.1362, simple_loss=0.1632, pruned_loss=0.05459, over 1937026.21 frames. ], batch size: 100, lr: 7.95e-03, grad_scale: 8.0 2022-12-07 22:13:16,820 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6987, 3.7907, 3.9088, 3.3684, 3.6922, 3.8845, 1.3796, 3.4715], device='cuda:1'), covar=tensor([0.0354, 0.0377, 0.0444, 0.0586, 0.0465, 0.0276, 0.3615, 0.0354], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0161, 0.0134, 0.0136, 0.0193, 0.0131, 0.0156, 0.0178], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:13:39,415 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5378, 3.8254, 3.5980, 3.4359, 2.8382, 3.8866, 3.5662, 1.9961], device='cuda:1'), covar=tensor([0.2404, 0.0720, 0.1126, 0.1268, 0.1046, 0.0732, 0.1351, 0.2500], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0072, 0.0059, 0.0061, 0.0087, 0.0070, 0.0092, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:1') 2022-12-07 22:13:40,284 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4569, 1.0152, 1.3646, 0.8710, 1.1441, 1.4039, 1.1224, 1.1687], device='cuda:1'), covar=tensor([0.0300, 0.0569, 0.0450, 0.0397, 0.0689, 0.0464, 0.0353, 0.1452], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0183, 0.0133, 0.0125, 0.0127, 0.0139, 0.0114, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:1') 2022-12-07 22:14:08,427 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8818, 2.5382, 2.6398, 1.6659, 2.4454, 2.5444, 2.8426, 2.3744], device='cuda:1'), covar=tensor([0.0668, 0.1016, 0.1003, 0.1860, 0.0938, 0.0916, 0.0668, 0.1417], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0184, 0.0134, 0.0127, 0.0128, 0.0140, 0.0115, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 22:14:22,585 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 2.276e+02 2.825e+02 3.448e+02 6.629e+02, threshold=5.650e+02, percent-clipped=5.0 2022-12-07 22:14:28,352 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1209, 1.0638, 0.9766, 0.9508, 1.0213, 0.4790, 0.8634, 1.0890], device='cuda:1'), covar=tensor([0.0290, 0.0499, 0.0308, 0.0396, 0.0259, 0.0315, 0.0859, 0.0460], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0026, 0.0023, 0.0025, 0.0035, 0.0025, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 22:14:33,334 INFO [train.py:873] (1/4) Epoch 10, batch 3500, loss[loss=0.1507, simple_loss=0.1514, pruned_loss=0.07503, over 2659.00 frames. ], tot_loss[loss=0.1364, simple_loss=0.163, pruned_loss=0.05488, over 1904398.54 frames. ], batch size: 100, lr: 7.95e-03, grad_scale: 8.0 2022-12-07 22:14:57,869 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6393, 5.4170, 4.8059, 5.1847, 5.1399, 5.4982, 5.5152, 5.5811], device='cuda:1'), covar=tensor([0.0481, 0.0280, 0.1967, 0.2000, 0.0588, 0.0512, 0.0836, 0.0644], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0244, 0.0412, 0.0528, 0.0306, 0.0399, 0.0377, 0.0344], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:15:21,677 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4753, 4.1403, 4.0802, 4.5094, 4.2085, 4.0303, 4.5330, 3.8250], device='cuda:1'), covar=tensor([0.0454, 0.1115, 0.0417, 0.0540, 0.0925, 0.0853, 0.0625, 0.0577], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0261, 0.0180, 0.0178, 0.0173, 0.0141, 0.0265, 0.0159], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 22:15:27,735 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:15:43,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 22:15:56,002 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-07 22:15:57,136 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.664e+02 3.016e+02 4.035e+02 8.090e+02, threshold=6.032e+02, percent-clipped=4.0 2022-12-07 22:15:57,352 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:16:02,308 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-07 22:16:07,711 INFO [train.py:873] (1/4) Epoch 10, batch 3600, loss[loss=0.2001, simple_loss=0.1704, pruned_loss=0.1149, over 1298.00 frames. ], tot_loss[loss=0.1353, simple_loss=0.1625, pruned_loss=0.054, over 1941345.70 frames. ], batch size: 100, lr: 7.94e-03, grad_scale: 8.0 2022-12-07 22:16:25,265 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:16:33,287 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:16:42,655 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:17:18,948 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:17:31,782 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.595e+01 2.418e+02 2.819e+02 3.742e+02 8.382e+02, threshold=5.637e+02, percent-clipped=6.0 2022-12-07 22:17:39,174 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7005, 2.3464, 3.8930, 3.9200, 3.8495, 2.3179, 3.8769, 3.0097], device='cuda:1'), covar=tensor([0.0320, 0.0765, 0.0537, 0.0339, 0.0286, 0.1199, 0.0240, 0.0729], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0247, 0.0365, 0.0313, 0.0255, 0.0298, 0.0282, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:17:42,531 INFO [train.py:873] (1/4) Epoch 10, batch 3700, loss[loss=0.1256, simple_loss=0.162, pruned_loss=0.04461, over 14446.00 frames. ], tot_loss[loss=0.1364, simple_loss=0.1633, pruned_loss=0.05473, over 1992464.63 frames. ], batch size: 51, lr: 7.94e-03, grad_scale: 8.0 2022-12-07 22:19:05,971 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 2.175e+02 2.639e+02 3.434e+02 5.313e+02, threshold=5.277e+02, percent-clipped=0.0 2022-12-07 22:19:16,068 INFO [train.py:873] (1/4) Epoch 10, batch 3800, loss[loss=0.1459, simple_loss=0.1685, pruned_loss=0.06159, over 10332.00 frames. ], tot_loss[loss=0.1369, simple_loss=0.1638, pruned_loss=0.05505, over 1959192.27 frames. ], batch size: 100, lr: 7.93e-03, grad_scale: 8.0 2022-12-07 22:19:30,077 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-07 22:19:59,554 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6271, 2.6498, 2.7769, 2.7674, 2.7252, 2.5170, 1.4518, 2.4331], device='cuda:1'), covar=tensor([0.0434, 0.0459, 0.0520, 0.0350, 0.0365, 0.0827, 0.2924, 0.0357], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0160, 0.0134, 0.0136, 0.0193, 0.0131, 0.0156, 0.0179], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:20:12,772 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:20:40,197 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.292e+02 2.855e+02 3.626e+02 5.894e+02, threshold=5.709e+02, percent-clipped=2.0 2022-12-07 22:20:49,793 INFO [train.py:873] (1/4) Epoch 10, batch 3900, loss[loss=0.1277, simple_loss=0.123, pruned_loss=0.06616, over 1202.00 frames. ], tot_loss[loss=0.1356, simple_loss=0.1628, pruned_loss=0.05416, over 1975635.94 frames. ], batch size: 100, lr: 7.93e-03, grad_scale: 4.0 2022-12-07 22:21:01,575 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:21:08,344 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8556, 3.6735, 3.3313, 3.4762, 3.7681, 3.7990, 3.8453, 3.8531], device='cuda:1'), covar=tensor([0.1019, 0.0557, 0.2297, 0.2891, 0.0808, 0.0852, 0.1098, 0.0913], device='cuda:1'), in_proj_covar=tensor([0.0358, 0.0246, 0.0418, 0.0534, 0.0313, 0.0403, 0.0381, 0.0346], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:21:09,306 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:21:55,987 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7217, 1.6657, 1.7017, 1.8110, 1.8695, 0.8665, 1.7908, 2.0088], device='cuda:1'), covar=tensor([0.0695, 0.1018, 0.0606, 0.1678, 0.1901, 0.0783, 0.1158, 0.0861], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0027, 0.0023, 0.0025, 0.0036, 0.0025, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 22:22:11,751 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-07 22:22:14,706 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.367e+02 2.840e+02 3.454e+02 7.877e+02, threshold=5.680e+02, percent-clipped=2.0 2022-12-07 22:22:23,934 INFO [train.py:873] (1/4) Epoch 10, batch 4000, loss[loss=0.1272, simple_loss=0.1631, pruned_loss=0.04567, over 14261.00 frames. ], tot_loss[loss=0.1347, simple_loss=0.1623, pruned_loss=0.05353, over 1988140.02 frames. ], batch size: 39, lr: 7.92e-03, grad_scale: 8.0 2022-12-07 22:22:31,718 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6564, 2.7350, 2.7015, 2.8008, 2.7458, 2.6365, 1.3413, 2.4354], device='cuda:1'), covar=tensor([0.0604, 0.0583, 0.0764, 0.0505, 0.0628, 0.0949, 0.3625, 0.0545], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0159, 0.0134, 0.0134, 0.0192, 0.0129, 0.0154, 0.0178], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:22:48,767 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6663, 1.5227, 4.1568, 3.9823, 3.9553, 4.2954, 3.5577, 4.2905], device='cuda:1'), covar=tensor([0.1737, 0.1680, 0.0139, 0.0214, 0.0224, 0.0128, 0.0234, 0.0144], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0122, 0.0166, 0.0142, 0.0132, 0.0114, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:22:50,572 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0525, 4.2231, 4.5205, 3.8863, 4.3833, 4.5063, 1.6767, 3.9850], device='cuda:1'), covar=tensor([0.0367, 0.0340, 0.0366, 0.0480, 0.0286, 0.0231, 0.3246, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0158, 0.0133, 0.0133, 0.0190, 0.0128, 0.0153, 0.0177], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:23:16,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4121, 1.9810, 2.4505, 2.4999, 2.2489, 1.9591, 2.4687, 2.1891], device='cuda:1'), covar=tensor([0.0242, 0.0521, 0.0228, 0.0178, 0.0319, 0.0591, 0.0218, 0.0346], device='cuda:1'), in_proj_covar=tensor([0.0274, 0.0245, 0.0363, 0.0310, 0.0253, 0.0297, 0.0282, 0.0275], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:23:29,169 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:23:48,079 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.453e+01 2.051e+02 2.497e+02 3.120e+02 6.082e+02, threshold=4.994e+02, percent-clipped=1.0 2022-12-07 22:23:57,371 INFO [train.py:873] (1/4) Epoch 10, batch 4100, loss[loss=0.1558, simple_loss=0.151, pruned_loss=0.08031, over 1246.00 frames. ], tot_loss[loss=0.1341, simple_loss=0.1618, pruned_loss=0.05318, over 1978939.40 frames. ], batch size: 100, lr: 7.91e-03, grad_scale: 4.0 2022-12-07 22:24:17,938 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2022-12-07 22:24:26,207 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:25:00,421 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2430, 3.9243, 3.7161, 3.9233, 4.0380, 4.1106, 4.2266, 4.2227], device='cuda:1'), covar=tensor([0.0791, 0.0573, 0.2131, 0.2230, 0.0782, 0.0737, 0.0914, 0.0734], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0246, 0.0417, 0.0532, 0.0313, 0.0401, 0.0380, 0.0348], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:25:22,726 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 2.564e+02 3.159e+02 3.830e+02 6.918e+02, threshold=6.317e+02, percent-clipped=10.0 2022-12-07 22:25:31,387 INFO [train.py:873] (1/4) Epoch 10, batch 4200, loss[loss=0.1122, simple_loss=0.1472, pruned_loss=0.03859, over 14019.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1626, pruned_loss=0.0541, over 1956744.90 frames. ], batch size: 19, lr: 7.91e-03, grad_scale: 4.0 2022-12-07 22:25:34,239 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0240, 2.9537, 3.9002, 2.6542, 2.3066, 3.1424, 1.7911, 3.0954], device='cuda:1'), covar=tensor([0.1352, 0.1182, 0.0540, 0.3122, 0.2541, 0.0941, 0.4154, 0.1325], device='cuda:1'), in_proj_covar=tensor([0.0078, 0.0093, 0.0087, 0.0093, 0.0112, 0.0079, 0.0125, 0.0084], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:1') 2022-12-07 22:25:44,052 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:25:46,917 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:25:59,780 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4632, 1.6784, 2.6973, 2.1291, 2.5663, 1.6819, 2.1752, 2.4074], device='cuda:1'), covar=tensor([0.1643, 0.3756, 0.0490, 0.3594, 0.1124, 0.3166, 0.1201, 0.0909], device='cuda:1'), in_proj_covar=tensor([0.0242, 0.0219, 0.0195, 0.0297, 0.0215, 0.0222, 0.0217, 0.0200], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:26:00,628 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3578, 3.9041, 2.9597, 4.8549, 4.3018, 4.5203, 3.7892, 3.2782], device='cuda:1'), covar=tensor([0.0959, 0.1408, 0.4359, 0.0486, 0.0902, 0.1377, 0.1378, 0.3810], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0302, 0.0275, 0.0248, 0.0305, 0.0293, 0.0260, 0.0260], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 22:26:16,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9736, 1.0028, 0.9512, 0.9273, 0.7819, 0.6042, 0.7504, 0.7413], device='cuda:1'), covar=tensor([0.0206, 0.0358, 0.0143, 0.0185, 0.0192, 0.0471, 0.0322, 0.0512], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0015, 0.0013, 0.0014, 0.0014, 0.0022, 0.0018, 0.0023], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 22:26:27,853 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:26:33,207 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:26:55,037 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.250e+02 2.962e+02 3.836e+02 8.638e+02, threshold=5.923e+02, percent-clipped=4.0 2022-12-07 22:26:58,068 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2022-12-07 22:27:03,694 INFO [train.py:873] (1/4) Epoch 10, batch 4300, loss[loss=0.117, simple_loss=0.1598, pruned_loss=0.03706, over 14002.00 frames. ], tot_loss[loss=0.1364, simple_loss=0.1632, pruned_loss=0.05475, over 1906874.92 frames. ], batch size: 26, lr: 7.90e-03, grad_scale: 4.0 2022-12-07 22:27:11,287 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 22:27:29,517 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:27:44,311 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7808, 3.5424, 3.4652, 3.8528, 3.3695, 3.1571, 3.8829, 3.7439], device='cuda:1'), covar=tensor([0.0616, 0.0938, 0.0857, 0.0572, 0.0925, 0.0784, 0.0617, 0.0713], device='cuda:1'), in_proj_covar=tensor([0.0125, 0.0119, 0.0129, 0.0137, 0.0130, 0.0108, 0.0152, 0.0128], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 22:28:17,191 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9814, 2.1971, 2.2109, 2.4115, 1.9247, 2.2509, 1.9873, 1.2569], device='cuda:1'), covar=tensor([0.1346, 0.0918, 0.1058, 0.0460, 0.1035, 0.0773, 0.1255, 0.2540], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0074, 0.0060, 0.0063, 0.0090, 0.0072, 0.0094, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:1') 2022-12-07 22:28:26,476 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.139e+02 2.736e+02 3.399e+02 6.791e+02, threshold=5.473e+02, percent-clipped=1.0 2022-12-07 22:28:34,601 INFO [train.py:873] (1/4) Epoch 10, batch 4400, loss[loss=0.1729, simple_loss=0.1512, pruned_loss=0.09727, over 1223.00 frames. ], tot_loss[loss=0.136, simple_loss=0.1632, pruned_loss=0.05444, over 1982094.46 frames. ], batch size: 100, lr: 7.90e-03, grad_scale: 8.0 2022-12-07 22:28:42,114 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6255, 2.4016, 3.1288, 1.9874, 1.9922, 2.6675, 1.3672, 2.5797], device='cuda:1'), covar=tensor([0.0950, 0.1461, 0.0554, 0.2082, 0.2284, 0.1101, 0.4268, 0.0940], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0095, 0.0088, 0.0094, 0.0112, 0.0079, 0.0126, 0.0085], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:1') 2022-12-07 22:28:58,312 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:29:52,100 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1178, 1.3763, 3.9822, 1.7049, 3.9838, 4.1346, 3.3579, 4.4966], device='cuda:1'), covar=tensor([0.0210, 0.3192, 0.0395, 0.2334, 0.0389, 0.0355, 0.0538, 0.0144], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0157, 0.0157, 0.0168, 0.0172, 0.0172, 0.0135, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 22:29:56,358 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.111e+02 2.675e+02 3.314e+02 5.959e+02, threshold=5.350e+02, percent-clipped=1.0 2022-12-07 22:30:05,238 INFO [train.py:873] (1/4) Epoch 10, batch 4500, loss[loss=0.1275, simple_loss=0.1612, pruned_loss=0.04691, over 14269.00 frames. ], tot_loss[loss=0.135, simple_loss=0.1625, pruned_loss=0.05375, over 1920408.37 frames. ], batch size: 46, lr: 7.89e-03, grad_scale: 8.0 2022-12-07 22:30:19,393 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:31:03,241 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:31:27,223 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.124e+02 2.389e+02 3.143e+02 3.960e+02 1.878e+03, threshold=6.285e+02, percent-clipped=10.0 2022-12-07 22:31:35,245 INFO [train.py:873] (1/4) Epoch 10, batch 4600, loss[loss=0.1272, simple_loss=0.1562, pruned_loss=0.04908, over 14040.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1628, pruned_loss=0.05395, over 1953475.03 frames. ], batch size: 19, lr: 7.89e-03, grad_scale: 8.0 2022-12-07 22:31:56,315 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:32:57,784 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.006e+02 2.571e+02 3.192e+02 6.224e+02, threshold=5.142e+02, percent-clipped=0.0 2022-12-07 22:33:06,364 INFO [train.py:873] (1/4) Epoch 10, batch 4700, loss[loss=0.1365, simple_loss=0.1681, pruned_loss=0.05245, over 14207.00 frames. ], tot_loss[loss=0.1351, simple_loss=0.1624, pruned_loss=0.05391, over 1947052.34 frames. ], batch size: 89, lr: 7.88e-03, grad_scale: 8.0 2022-12-07 22:33:29,643 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:33:55,147 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2022-12-07 22:34:12,939 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:34:27,384 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.194e+02 2.219e+02 2.922e+02 3.522e+02 5.753e+02, threshold=5.843e+02, percent-clipped=6.0 2022-12-07 22:34:27,565 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7811, 0.7462, 0.6442, 0.7882, 0.8446, 0.2643, 0.7219, 0.8728], device='cuda:1'), covar=tensor([0.0261, 0.0440, 0.0355, 0.0342, 0.0242, 0.0281, 0.0583, 0.0469], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0027, 0.0024, 0.0025, 0.0037, 0.0025, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 22:34:35,285 INFO [train.py:873] (1/4) Epoch 10, batch 4800, loss[loss=0.2092, simple_loss=0.1735, pruned_loss=0.1225, over 1221.00 frames. ], tot_loss[loss=0.1362, simple_loss=0.1628, pruned_loss=0.05476, over 1934870.90 frames. ], batch size: 100, lr: 7.88e-03, grad_scale: 8.0 2022-12-07 22:34:44,718 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4145, 4.8475, 4.8388, 5.3675, 4.9246, 4.5109, 5.3634, 4.3468], device='cuda:1'), covar=tensor([0.0304, 0.1066, 0.0380, 0.0410, 0.0826, 0.0436, 0.0505, 0.0558], device='cuda:1'), in_proj_covar=tensor([0.0162, 0.0258, 0.0178, 0.0175, 0.0171, 0.0141, 0.0266, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 22:35:06,551 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:35:56,441 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 2.330e+02 2.761e+02 3.317e+02 8.665e+02, threshold=5.521e+02, percent-clipped=2.0 2022-12-07 22:36:01,285 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:36:04,982 INFO [train.py:873] (1/4) Epoch 10, batch 4900, loss[loss=0.1448, simple_loss=0.1648, pruned_loss=0.06241, over 5961.00 frames. ], tot_loss[loss=0.1367, simple_loss=0.1637, pruned_loss=0.0548, over 2020011.85 frames. ], batch size: 100, lr: 7.87e-03, grad_scale: 8.0 2022-12-07 22:36:09,388 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 22:36:24,931 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:37:03,934 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 22:37:04,739 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0301, 2.7643, 3.8097, 2.6104, 2.2256, 3.1598, 1.7261, 3.0105], device='cuda:1'), covar=tensor([0.1183, 0.1239, 0.0568, 0.2045, 0.2351, 0.0862, 0.3746, 0.0951], device='cuda:1'), in_proj_covar=tensor([0.0078, 0.0094, 0.0087, 0.0092, 0.0111, 0.0077, 0.0123, 0.0084], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:1') 2022-12-07 22:37:08,118 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:37:26,027 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.275e+02 2.705e+02 3.390e+02 8.104e+02, threshold=5.410e+02, percent-clipped=1.0 2022-12-07 22:37:33,743 INFO [train.py:873] (1/4) Epoch 10, batch 5000, loss[loss=0.1387, simple_loss=0.1695, pruned_loss=0.05393, over 14452.00 frames. ], tot_loss[loss=0.1347, simple_loss=0.1628, pruned_loss=0.05331, over 2079689.66 frames. ], batch size: 49, lr: 7.87e-03, grad_scale: 8.0 2022-12-07 22:37:41,914 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9797, 3.2764, 2.5422, 4.2493, 3.9432, 4.1139, 3.2112, 2.7022], device='cuda:1'), covar=tensor([0.1247, 0.2030, 0.5392, 0.0750, 0.1139, 0.1403, 0.2012, 0.4557], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0298, 0.0271, 0.0248, 0.0304, 0.0294, 0.0255, 0.0259], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 22:38:06,606 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8282, 1.9430, 2.7992, 2.2305, 2.7923, 2.6079, 2.4674, 2.3429], device='cuda:1'), covar=tensor([0.0675, 0.2831, 0.0916, 0.1961, 0.0586, 0.1197, 0.0946, 0.1625], device='cuda:1'), in_proj_covar=tensor([0.0341, 0.0315, 0.0398, 0.0303, 0.0374, 0.0318, 0.0361, 0.0314], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:38:22,502 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8830, 1.6323, 2.1327, 1.7263, 1.9627, 1.4978, 1.7065, 1.9975], device='cuda:1'), covar=tensor([0.1829, 0.2342, 0.0286, 0.1433, 0.0905, 0.1142, 0.0752, 0.0424], device='cuda:1'), in_proj_covar=tensor([0.0242, 0.0218, 0.0194, 0.0295, 0.0212, 0.0221, 0.0216, 0.0199], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:38:54,056 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.200e+01 2.088e+02 2.701e+02 3.421e+02 5.931e+02, threshold=5.402e+02, percent-clipped=1.0 2022-12-07 22:39:01,815 INFO [train.py:873] (1/4) Epoch 10, batch 5100, loss[loss=0.1347, simple_loss=0.1613, pruned_loss=0.05402, over 14249.00 frames. ], tot_loss[loss=0.1345, simple_loss=0.1621, pruned_loss=0.05345, over 2065004.98 frames. ], batch size: 63, lr: 7.86e-03, grad_scale: 4.0 2022-12-07 22:39:47,546 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 22:40:21,983 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:40:22,701 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.094e+02 2.487e+02 3.025e+02 3.884e+02 1.180e+03, threshold=6.049e+02, percent-clipped=5.0 2022-12-07 22:40:28,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.56 vs. limit=2.0 2022-12-07 22:40:29,836 INFO [train.py:873] (1/4) Epoch 10, batch 5200, loss[loss=0.1122, simple_loss=0.1517, pruned_loss=0.03636, over 13554.00 frames. ], tot_loss[loss=0.1347, simple_loss=0.1626, pruned_loss=0.05343, over 2040444.65 frames. ], batch size: 100, lr: 7.85e-03, grad_scale: 8.0 2022-12-07 22:40:53,524 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8838, 1.7228, 2.0160, 1.7459, 2.0555, 1.8728, 1.6792, 1.9027], device='cuda:1'), covar=tensor([0.0606, 0.1123, 0.0229, 0.0287, 0.0250, 0.0675, 0.0218, 0.0283], device='cuda:1'), in_proj_covar=tensor([0.0340, 0.0316, 0.0399, 0.0305, 0.0373, 0.0317, 0.0362, 0.0314], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:40:57,395 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-07 22:41:00,525 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:41:25,164 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 22:41:51,221 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3274, 2.0904, 2.6952, 1.7118, 1.8377, 2.3421, 1.2715, 2.4086], device='cuda:1'), covar=tensor([0.1015, 0.2079, 0.0644, 0.2442, 0.2443, 0.0982, 0.4288, 0.1002], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0094, 0.0087, 0.0093, 0.0112, 0.0078, 0.0124, 0.0086], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:41:52,767 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.362e+02 2.946e+02 3.664e+02 8.798e+02, threshold=5.892e+02, percent-clipped=2.0 2022-12-07 22:41:54,615 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:41:59,540 INFO [train.py:873] (1/4) Epoch 10, batch 5300, loss[loss=0.1928, simple_loss=0.1645, pruned_loss=0.1106, over 1297.00 frames. ], tot_loss[loss=0.1352, simple_loss=0.1627, pruned_loss=0.05382, over 2017826.70 frames. ], batch size: 100, lr: 7.85e-03, grad_scale: 4.0 2022-12-07 22:42:41,744 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5654, 1.3993, 3.4995, 1.6290, 3.4568, 3.5492, 2.5638, 3.8743], device='cuda:1'), covar=tensor([0.0218, 0.3036, 0.0411, 0.2144, 0.0667, 0.0375, 0.0842, 0.0182], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0159, 0.0158, 0.0170, 0.0172, 0.0172, 0.0134, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 22:42:44,486 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.60 vs. limit=5.0 2022-12-07 22:43:21,802 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 2.215e+02 2.822e+02 3.427e+02 6.404e+02, threshold=5.643e+02, percent-clipped=2.0 2022-12-07 22:43:27,863 INFO [train.py:873] (1/4) Epoch 10, batch 5400, loss[loss=0.1834, simple_loss=0.1933, pruned_loss=0.08677, over 10345.00 frames. ], tot_loss[loss=0.1348, simple_loss=0.1626, pruned_loss=0.05351, over 2067074.99 frames. ], batch size: 100, lr: 7.84e-03, grad_scale: 4.0 2022-12-07 22:43:51,894 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8548, 1.6963, 1.9689, 1.6065, 2.0076, 1.8412, 1.6206, 1.8253], device='cuda:1'), covar=tensor([0.0613, 0.0895, 0.0254, 0.0414, 0.0306, 0.0584, 0.0256, 0.0326], device='cuda:1'), in_proj_covar=tensor([0.0340, 0.0316, 0.0400, 0.0304, 0.0374, 0.0314, 0.0363, 0.0314], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:44:34,085 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8703, 1.6457, 1.9795, 1.6915, 2.0404, 1.8174, 1.6760, 1.8993], device='cuda:1'), covar=tensor([0.0536, 0.1319, 0.0293, 0.0418, 0.0328, 0.0742, 0.0294, 0.0333], device='cuda:1'), in_proj_covar=tensor([0.0340, 0.0316, 0.0398, 0.0304, 0.0373, 0.0313, 0.0362, 0.0313], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:44:48,875 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:44:50,661 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.129e+02 2.539e+02 3.119e+02 4.865e+02, threshold=5.078e+02, percent-clipped=0.0 2022-12-07 22:44:57,304 INFO [train.py:873] (1/4) Epoch 10, batch 5500, loss[loss=0.1458, simple_loss=0.1344, pruned_loss=0.07864, over 2613.00 frames. ], tot_loss[loss=0.1334, simple_loss=0.1615, pruned_loss=0.05265, over 2034809.98 frames. ], batch size: 100, lr: 7.84e-03, grad_scale: 4.0 2022-12-07 22:44:58,510 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-07 22:45:12,475 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 22:45:24,733 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:45:31,972 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:45:50,899 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 22:45:54,243 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1081, 1.3509, 3.2151, 1.5321, 3.1170, 3.2484, 2.2516, 3.3957], device='cuda:1'), covar=tensor([0.0251, 0.2984, 0.0347, 0.2204, 0.0934, 0.0357, 0.0975, 0.0212], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0158, 0.0156, 0.0169, 0.0170, 0.0170, 0.0132, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 22:46:04,746 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3757, 2.1657, 3.2988, 3.4150, 3.3244, 2.2719, 3.2390, 2.6155], device='cuda:1'), covar=tensor([0.0284, 0.0704, 0.0526, 0.0334, 0.0304, 0.1022, 0.0288, 0.0719], device='cuda:1'), in_proj_covar=tensor([0.0269, 0.0242, 0.0359, 0.0309, 0.0251, 0.0290, 0.0282, 0.0271], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:46:17,384 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:46:18,281 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8261, 3.4473, 2.7738, 4.1580, 3.9448, 3.8802, 3.4333, 2.7399], device='cuda:1'), covar=tensor([0.1158, 0.1625, 0.4116, 0.0497, 0.0840, 0.1774, 0.1378, 0.3911], device='cuda:1'), in_proj_covar=tensor([0.0267, 0.0295, 0.0268, 0.0243, 0.0301, 0.0292, 0.0254, 0.0256], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 22:46:19,130 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:46:19,731 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.890e+01 2.291e+02 2.969e+02 3.792e+02 1.060e+03, threshold=5.938e+02, percent-clipped=6.0 2022-12-07 22:46:25,680 INFO [train.py:873] (1/4) Epoch 10, batch 5600, loss[loss=0.1611, simple_loss=0.1527, pruned_loss=0.08479, over 1308.00 frames. ], tot_loss[loss=0.1347, simple_loss=0.162, pruned_loss=0.05366, over 1993636.56 frames. ], batch size: 100, lr: 7.83e-03, grad_scale: 8.0 2022-12-07 22:46:33,870 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 22:47:30,702 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:47:48,144 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.591e+02 3.035e+02 3.423e+02 1.031e+03, threshold=6.069e+02, percent-clipped=1.0 2022-12-07 22:47:55,242 INFO [train.py:873] (1/4) Epoch 10, batch 5700, loss[loss=0.1205, simple_loss=0.1451, pruned_loss=0.04798, over 4952.00 frames. ], tot_loss[loss=0.1345, simple_loss=0.1621, pruned_loss=0.05343, over 2006939.25 frames. ], batch size: 100, lr: 7.83e-03, grad_scale: 8.0 2022-12-07 22:48:24,509 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:49:17,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.156e+02 2.743e+02 3.331e+02 7.800e+02, threshold=5.486e+02, percent-clipped=2.0 2022-12-07 22:49:17,748 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-07 22:49:23,295 INFO [train.py:873] (1/4) Epoch 10, batch 5800, loss[loss=0.1612, simple_loss=0.1463, pruned_loss=0.08805, over 1164.00 frames. ], tot_loss[loss=0.1341, simple_loss=0.1617, pruned_loss=0.05329, over 1985817.86 frames. ], batch size: 100, lr: 7.82e-03, grad_scale: 8.0 2022-12-07 22:49:27,348 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-07 22:50:09,385 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 22:50:20,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8561, 1.3359, 2.9500, 2.6580, 2.8284, 2.9956, 2.0905, 2.9315], device='cuda:1'), covar=tensor([0.1059, 0.1322, 0.0123, 0.0299, 0.0296, 0.0130, 0.0427, 0.0156], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0122, 0.0166, 0.0142, 0.0133, 0.0117, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 22:50:40,931 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:50:44,054 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:50:46,442 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.192e+02 2.728e+02 3.266e+02 1.115e+03, threshold=5.456e+02, percent-clipped=1.0 2022-12-07 22:50:52,925 INFO [train.py:873] (1/4) Epoch 10, batch 5900, loss[loss=0.1558, simple_loss=0.1792, pruned_loss=0.06623, over 11231.00 frames. ], tot_loss[loss=0.1338, simple_loss=0.1612, pruned_loss=0.05322, over 1939651.98 frames. ], batch size: 100, lr: 7.82e-03, grad_scale: 8.0 2022-12-07 22:50:53,128 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4324, 3.2486, 3.0751, 2.1054, 3.0088, 3.2381, 3.6862, 2.8138], device='cuda:1'), covar=tensor([0.0635, 0.1246, 0.1176, 0.2065, 0.0932, 0.0794, 0.0498, 0.1440], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0185, 0.0134, 0.0126, 0.0133, 0.0139, 0.0114, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 22:50:54,743 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 22:51:17,547 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3035, 3.6115, 4.1973, 3.1514, 2.7109, 3.1603, 2.1570, 3.4446], device='cuda:1'), covar=tensor([0.1326, 0.0942, 0.0765, 0.1590, 0.2329, 0.1161, 0.3889, 0.0931], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0094, 0.0088, 0.0094, 0.0113, 0.0079, 0.0126, 0.0087], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:1') 2022-12-07 22:51:26,772 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:51:58,680 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1486, 1.9781, 1.8296, 1.8642, 2.0642, 2.0575, 2.0946, 2.0637], device='cuda:1'), covar=tensor([0.1210, 0.1365, 0.2843, 0.3102, 0.1504, 0.1494, 0.1949, 0.1239], device='cuda:1'), in_proj_covar=tensor([0.0357, 0.0253, 0.0418, 0.0551, 0.0318, 0.0409, 0.0384, 0.0352], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:52:13,163 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.43 vs. limit=5.0 2022-12-07 22:52:14,955 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.132e+02 2.894e+02 3.484e+02 7.158e+02, threshold=5.787e+02, percent-clipped=1.0 2022-12-07 22:52:20,977 INFO [train.py:873] (1/4) Epoch 10, batch 6000, loss[loss=0.1428, simple_loss=0.1749, pruned_loss=0.05541, over 14296.00 frames. ], tot_loss[loss=0.1334, simple_loss=0.1612, pruned_loss=0.05274, over 1950152.74 frames. ], batch size: 25, lr: 7.81e-03, grad_scale: 8.0 2022-12-07 22:52:20,977 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 22:52:36,795 INFO [train.py:905] (1/4) Epoch 10, validation: loss=0.1251, simple_loss=0.167, pruned_loss=0.04163, over 857387.00 frames. 2022-12-07 22:52:36,796 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 22:52:42,010 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:53:02,909 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:53:16,079 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4480, 1.4689, 1.6348, 1.4519, 1.3553, 1.2133, 1.0521, 0.9733], device='cuda:1'), covar=tensor([0.0300, 0.0381, 0.0277, 0.0293, 0.0272, 0.0321, 0.0281, 0.0470], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0015, 0.0013, 0.0014, 0.0014, 0.0023, 0.0018, 0.0023], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 22:53:35,930 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:54:00,209 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.199e+02 2.430e+02 2.883e+02 3.808e+02 1.343e+03, threshold=5.765e+02, percent-clipped=6.0 2022-12-07 22:54:06,422 INFO [train.py:873] (1/4) Epoch 10, batch 6100, loss[loss=0.129, simple_loss=0.1292, pruned_loss=0.06445, over 2618.00 frames. ], tot_loss[loss=0.1336, simple_loss=0.1618, pruned_loss=0.05267, over 1994835.67 frames. ], batch size: 100, lr: 7.81e-03, grad_scale: 8.0 2022-12-07 22:54:07,757 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-07 22:54:14,616 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.56 vs. limit=5.0 2022-12-07 22:55:10,647 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0604, 1.8786, 3.2721, 2.4037, 3.0864, 1.8309, 2.5291, 3.0606], device='cuda:1'), covar=tensor([0.0809, 0.4442, 0.0451, 0.5529, 0.0741, 0.3691, 0.1263, 0.0548], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0219, 0.0195, 0.0299, 0.0217, 0.0224, 0.0220, 0.0202], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 22:55:16,413 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7196, 1.6020, 1.6398, 1.5744, 1.6930, 0.8543, 1.6002, 1.9084], device='cuda:1'), covar=tensor([0.0859, 0.1130, 0.1316, 0.1783, 0.1654, 0.0970, 0.0926, 0.0917], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0025, 0.0027, 0.0024, 0.0025, 0.0037, 0.0025, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 22:55:18,958 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:55:23,975 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:55:29,018 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.250e+02 2.648e+02 3.376e+02 5.030e+02, threshold=5.296e+02, percent-clipped=0.0 2022-12-07 22:55:35,296 INFO [train.py:873] (1/4) Epoch 10, batch 6200, loss[loss=0.1407, simple_loss=0.1694, pruned_loss=0.05597, over 14413.00 frames. ], tot_loss[loss=0.1344, simple_loss=0.162, pruned_loss=0.05342, over 1950841.76 frames. ], batch size: 41, lr: 7.80e-03, grad_scale: 8.0 2022-12-07 22:56:06,620 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:56:13,670 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:56:58,218 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.014e+02 2.284e+02 2.758e+02 3.409e+02 5.517e+02, threshold=5.515e+02, percent-clipped=2.0 2022-12-07 22:57:04,251 INFO [train.py:873] (1/4) Epoch 10, batch 6300, loss[loss=0.139, simple_loss=0.1662, pruned_loss=0.05587, over 14107.00 frames. ], tot_loss[loss=0.1333, simple_loss=0.1611, pruned_loss=0.05275, over 1920176.30 frames. ], batch size: 19, lr: 7.80e-03, grad_scale: 8.0 2022-12-07 22:57:30,142 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:57:57,835 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:58:03,455 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9580, 2.7064, 2.7521, 2.9312, 2.7948, 2.8809, 2.9598, 2.4700], device='cuda:1'), covar=tensor([0.0509, 0.1100, 0.0549, 0.0551, 0.0790, 0.0534, 0.0716, 0.0633], device='cuda:1'), in_proj_covar=tensor([0.0158, 0.0255, 0.0177, 0.0176, 0.0170, 0.0142, 0.0263, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 22:58:11,964 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:58:17,929 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8742, 1.5202, 1.7881, 2.0495, 1.4311, 1.8466, 1.7745, 1.9489], device='cuda:1'), covar=tensor([0.0100, 0.0169, 0.0099, 0.0082, 0.0205, 0.0191, 0.0134, 0.0087], device='cuda:1'), in_proj_covar=tensor([0.0271, 0.0241, 0.0361, 0.0307, 0.0249, 0.0289, 0.0282, 0.0270], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 22:58:25,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.328e+02 2.936e+02 3.672e+02 8.689e+02, threshold=5.873e+02, percent-clipped=4.0 2022-12-07 22:58:32,158 INFO [train.py:873] (1/4) Epoch 10, batch 6400, loss[loss=0.1295, simple_loss=0.1634, pruned_loss=0.04785, over 14254.00 frames. ], tot_loss[loss=0.1332, simple_loss=0.1612, pruned_loss=0.05258, over 1925956.34 frames. ], batch size: 80, lr: 7.79e-03, grad_scale: 8.0 2022-12-07 22:59:21,295 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3824, 2.3723, 2.5403, 2.5178, 2.5027, 2.0545, 1.4745, 2.2494], device='cuda:1'), covar=tensor([0.0518, 0.0420, 0.0410, 0.0382, 0.0451, 0.1195, 0.2178, 0.0405], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0161, 0.0135, 0.0134, 0.0193, 0.0130, 0.0155, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 22:59:53,824 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.247e+02 2.848e+02 3.548e+02 7.995e+02, threshold=5.696e+02, percent-clipped=2.0 2022-12-07 22:59:59,601 INFO [train.py:873] (1/4) Epoch 10, batch 6500, loss[loss=0.1393, simple_loss=0.1485, pruned_loss=0.06502, over 3865.00 frames. ], tot_loss[loss=0.1337, simple_loss=0.1614, pruned_loss=0.05299, over 1917138.12 frames. ], batch size: 100, lr: 7.79e-03, grad_scale: 8.0 2022-12-07 23:00:20,983 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3842, 4.9871, 4.7860, 5.4100, 4.9734, 4.4003, 5.3707, 5.3660], device='cuda:1'), covar=tensor([0.0550, 0.0676, 0.0792, 0.0393, 0.0668, 0.0561, 0.0578, 0.0470], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0122, 0.0130, 0.0139, 0.0134, 0.0110, 0.0155, 0.0131], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:00:31,631 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:00:32,458 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:01:15,008 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-07 23:01:22,223 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 2.214e+02 2.798e+02 3.905e+02 7.921e+02, threshold=5.597e+02, percent-clipped=3.0 2022-12-07 23:01:25,171 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:01:28,052 INFO [train.py:873] (1/4) Epoch 10, batch 6600, loss[loss=0.09928, simple_loss=0.1402, pruned_loss=0.02918, over 14376.00 frames. ], tot_loss[loss=0.1327, simple_loss=0.1608, pruned_loss=0.05229, over 1956678.43 frames. ], batch size: 41, lr: 7.78e-03, grad_scale: 4.0 2022-12-07 23:01:39,779 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 23:01:48,002 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 23:02:00,308 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 23:02:00,639 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6158, 1.5091, 3.6061, 1.5039, 3.5545, 3.7533, 2.4946, 3.9717], device='cuda:1'), covar=tensor([0.0253, 0.3175, 0.0417, 0.2490, 0.0568, 0.0373, 0.0768, 0.0188], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0160, 0.0159, 0.0170, 0.0173, 0.0174, 0.0135, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:02:22,383 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:02:51,954 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.183e+02 2.052e+02 2.533e+02 3.147e+02 6.445e+02, threshold=5.065e+02, percent-clipped=3.0 2022-12-07 23:02:57,097 INFO [train.py:873] (1/4) Epoch 10, batch 6700, loss[loss=0.1261, simple_loss=0.1545, pruned_loss=0.04881, over 12802.00 frames. ], tot_loss[loss=0.1331, simple_loss=0.1612, pruned_loss=0.05256, over 1948353.96 frames. ], batch size: 100, lr: 7.78e-03, grad_scale: 4.0 2022-12-07 23:03:03,569 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 23:03:04,993 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:03:12,337 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3723, 3.0249, 2.4152, 3.4902, 3.2508, 3.3237, 2.8545, 2.3675], device='cuda:1'), covar=tensor([0.0896, 0.1623, 0.3440, 0.0576, 0.1036, 0.1103, 0.1412, 0.3706], device='cuda:1'), in_proj_covar=tensor([0.0264, 0.0296, 0.0266, 0.0246, 0.0303, 0.0291, 0.0254, 0.0255], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:03:12,807 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-07 23:03:32,886 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7164, 2.7420, 2.5570, 2.8912, 2.4361, 2.5580, 2.7721, 2.7752], device='cuda:1'), covar=tensor([0.0760, 0.0990, 0.0938, 0.0715, 0.0993, 0.0857, 0.0906, 0.0854], device='cuda:1'), in_proj_covar=tensor([0.0126, 0.0122, 0.0131, 0.0140, 0.0134, 0.0111, 0.0157, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:03:40,726 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-07 23:03:53,724 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3671, 1.4541, 2.5306, 1.4048, 2.5101, 2.5100, 1.8535, 2.6241], device='cuda:1'), covar=tensor([0.0249, 0.2150, 0.0321, 0.1702, 0.0455, 0.0465, 0.0991, 0.0252], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0159, 0.0157, 0.0168, 0.0171, 0.0173, 0.0133, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:03:57,611 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 23:04:19,047 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.393e+02 2.874e+02 3.943e+02 8.045e+02, threshold=5.748e+02, percent-clipped=8.0 2022-12-07 23:04:21,936 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4885, 3.5974, 3.7351, 3.3419, 3.6369, 3.4537, 1.5525, 3.4392], device='cuda:1'), covar=tensor([0.0304, 0.0321, 0.0372, 0.0493, 0.0328, 0.0455, 0.3072, 0.0295], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0161, 0.0136, 0.0135, 0.0192, 0.0131, 0.0155, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:04:24,812 INFO [train.py:873] (1/4) Epoch 10, batch 6800, loss[loss=0.1413, simple_loss=0.1337, pruned_loss=0.0744, over 1225.00 frames. ], tot_loss[loss=0.1336, simple_loss=0.1615, pruned_loss=0.05284, over 1981231.74 frames. ], batch size: 100, lr: 7.77e-03, grad_scale: 8.0 2022-12-07 23:04:56,567 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:08,671 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:38,784 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:38,957 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9348, 2.1048, 3.0056, 3.0488, 3.0371, 2.1714, 2.8582, 2.3425], device='cuda:1'), covar=tensor([0.0293, 0.0683, 0.0509, 0.0317, 0.0299, 0.0939, 0.0280, 0.0652], device='cuda:1'), in_proj_covar=tensor([0.0271, 0.0243, 0.0362, 0.0307, 0.0251, 0.0292, 0.0282, 0.0270], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:05:45,391 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:47,781 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 2.283e+02 2.713e+02 3.632e+02 6.989e+02, threshold=5.426e+02, percent-clipped=3.0 2022-12-07 23:05:52,206 INFO [train.py:873] (1/4) Epoch 10, batch 6900, loss[loss=0.1383, simple_loss=0.1682, pruned_loss=0.05426, over 14283.00 frames. ], tot_loss[loss=0.1345, simple_loss=0.1619, pruned_loss=0.05355, over 1922406.66 frames. ], batch size: 69, lr: 7.77e-03, grad_scale: 4.0 2022-12-07 23:05:56,660 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:06:01,556 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:06:02,007 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2022-12-07 23:06:42,358 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-07 23:06:52,428 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:07:16,832 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.278e+02 2.897e+02 3.473e+02 8.232e+02, threshold=5.794e+02, percent-clipped=1.0 2022-12-07 23:07:21,370 INFO [train.py:873] (1/4) Epoch 10, batch 7000, loss[loss=0.1399, simple_loss=0.1612, pruned_loss=0.0593, over 7798.00 frames. ], tot_loss[loss=0.1346, simple_loss=0.1622, pruned_loss=0.05355, over 1945518.19 frames. ], batch size: 100, lr: 7.76e-03, grad_scale: 4.0 2022-12-07 23:07:51,075 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2022-12-07 23:08:45,730 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.185e+02 2.832e+02 3.460e+02 7.150e+02, threshold=5.664e+02, percent-clipped=2.0 2022-12-07 23:08:49,892 INFO [train.py:873] (1/4) Epoch 10, batch 7100, loss[loss=0.1424, simple_loss=0.1754, pruned_loss=0.05474, over 14487.00 frames. ], tot_loss[loss=0.1333, simple_loss=0.1613, pruned_loss=0.05262, over 1983642.08 frames. ], batch size: 49, lr: 7.76e-03, grad_scale: 4.0 2022-12-07 23:08:57,619 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6430, 2.2761, 2.9619, 1.8117, 1.7599, 2.5502, 1.3086, 2.5619], device='cuda:1'), covar=tensor([0.0862, 0.1639, 0.0639, 0.2605, 0.3200, 0.0963, 0.4583, 0.1252], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0094, 0.0089, 0.0095, 0.0114, 0.0080, 0.0126, 0.0087], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:1') 2022-12-07 23:09:09,078 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1665, 1.9879, 2.0235, 2.1368, 2.0513, 1.9750, 2.1998, 1.8423], device='cuda:1'), covar=tensor([0.0674, 0.1269, 0.0688, 0.0824, 0.1040, 0.0707, 0.0884, 0.0723], device='cuda:1'), in_proj_covar=tensor([0.0160, 0.0256, 0.0179, 0.0174, 0.0174, 0.0141, 0.0262, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:09:25,641 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.27 vs. limit=5.0 2022-12-07 23:09:41,753 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.23 vs. limit=5.0 2022-12-07 23:09:51,412 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:09:52,449 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2022-12-07 23:10:11,105 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:13,566 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.322e+02 2.903e+02 3.521e+02 5.847e+02, threshold=5.807e+02, percent-clipped=2.0 2022-12-07 23:10:18,336 INFO [train.py:873] (1/4) Epoch 10, batch 7200, loss[loss=0.1549, simple_loss=0.1766, pruned_loss=0.06663, over 12745.00 frames. ], tot_loss[loss=0.1337, simple_loss=0.1615, pruned_loss=0.05299, over 1979536.08 frames. ], batch size: 100, lr: 7.75e-03, grad_scale: 8.0 2022-12-07 23:10:23,654 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:28,216 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:43,833 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8823, 2.3666, 3.3710, 2.1744, 2.0300, 2.8528, 1.5918, 2.7249], device='cuda:1'), covar=tensor([0.0728, 0.1290, 0.0539, 0.1817, 0.2437, 0.0987, 0.3921, 0.1314], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0093, 0.0088, 0.0094, 0.0113, 0.0079, 0.0124, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:1') 2022-12-07 23:10:44,739 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:53,492 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:11:11,903 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:11:21,438 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:11:41,652 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 2.194e+02 2.882e+02 3.919e+02 8.679e+02, threshold=5.764e+02, percent-clipped=4.0 2022-12-07 23:11:46,062 INFO [train.py:873] (1/4) Epoch 10, batch 7300, loss[loss=0.08948, simple_loss=0.1307, pruned_loss=0.02413, over 13889.00 frames. ], tot_loss[loss=0.1325, simple_loss=0.1604, pruned_loss=0.05229, over 2024011.33 frames. ], batch size: 20, lr: 7.75e-03, grad_scale: 8.0 2022-12-07 23:13:09,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.421e+02 2.835e+02 3.319e+02 6.698e+02, threshold=5.670e+02, percent-clipped=3.0 2022-12-07 23:13:14,008 INFO [train.py:873] (1/4) Epoch 10, batch 7400, loss[loss=0.1717, simple_loss=0.1833, pruned_loss=0.08009, over 9485.00 frames. ], tot_loss[loss=0.1329, simple_loss=0.1606, pruned_loss=0.05258, over 2003521.08 frames. ], batch size: 100, lr: 7.74e-03, grad_scale: 8.0 2022-12-07 23:13:32,778 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 23:13:49,264 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-07 23:13:58,889 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6635, 3.3091, 2.5593, 3.8644, 3.6429, 3.6406, 3.2447, 2.5415], device='cuda:1'), covar=tensor([0.0904, 0.1584, 0.3722, 0.0550, 0.0977, 0.1426, 0.1195, 0.4080], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0297, 0.0267, 0.0246, 0.0306, 0.0291, 0.0252, 0.0257], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:14:06,332 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 23:14:06,951 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2022-12-07 23:14:23,291 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-07 23:14:38,882 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.214e+02 2.894e+02 3.789e+02 9.059e+02, threshold=5.789e+02, percent-clipped=5.0 2022-12-07 23:14:42,300 INFO [train.py:873] (1/4) Epoch 10, batch 7500, loss[loss=0.1477, simple_loss=0.1647, pruned_loss=0.06533, over 11948.00 frames. ], tot_loss[loss=0.1335, simple_loss=0.1611, pruned_loss=0.05295, over 1996678.08 frames. ], batch size: 100, lr: 7.73e-03, grad_scale: 4.0 2022-12-07 23:14:47,604 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:14:50,127 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0571, 1.8084, 4.6936, 4.2511, 4.2335, 4.7968, 4.4932, 4.7676], device='cuda:1'), covar=tensor([0.1367, 0.1418, 0.0082, 0.0165, 0.0199, 0.0091, 0.0121, 0.0103], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0162, 0.0126, 0.0170, 0.0144, 0.0137, 0.0119, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:15:04,157 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:15:04,744 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-07 23:15:24,290 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:16:08,852 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:16:09,627 INFO [train.py:873] (1/4) Epoch 11, batch 0, loss[loss=0.1352, simple_loss=0.1506, pruned_loss=0.05989, over 3908.00 frames. ], tot_loss[loss=0.1352, simple_loss=0.1506, pruned_loss=0.05989, over 3908.00 frames. ], batch size: 100, lr: 7.38e-03, grad_scale: 8.0 2022-12-07 23:16:09,627 INFO [train.py:896] (1/4) Computing validation loss 2022-12-07 23:16:16,873 INFO [train.py:905] (1/4) Epoch 11, validation: loss=0.1341, simple_loss=0.1756, pruned_loss=0.0463, over 857387.00 frames. 2022-12-07 23:16:16,873 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-07 23:16:21,386 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:16:47,137 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.744e+01 1.774e+02 2.985e+02 4.009e+02 1.076e+03, threshold=5.970e+02, percent-clipped=9.0 2022-12-07 23:16:58,373 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:17:01,182 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4576, 1.4117, 1.5512, 1.2842, 1.5049, 0.8434, 1.4684, 1.6565], device='cuda:1'), covar=tensor([0.1377, 0.1118, 0.1336, 0.0802, 0.0774, 0.1192, 0.1171, 0.0738], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0027, 0.0024, 0.0025, 0.0037, 0.0025, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:17:30,776 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 23:17:36,645 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-07 23:17:45,591 INFO [train.py:873] (1/4) Epoch 11, batch 100, loss[loss=0.1216, simple_loss=0.1565, pruned_loss=0.04335, over 14314.00 frames. ], tot_loss[loss=0.1328, simple_loss=0.1621, pruned_loss=0.05173, over 876090.98 frames. ], batch size: 28, lr: 7.38e-03, grad_scale: 8.0 2022-12-07 23:18:14,580 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.145e+02 2.643e+02 3.199e+02 5.049e+02, threshold=5.286e+02, percent-clipped=0.0 2022-12-07 23:18:22,452 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5619, 1.5289, 1.8903, 1.4798, 1.5739, 0.8465, 1.6678, 1.8576], device='cuda:1'), covar=tensor([0.1380, 0.1217, 0.0614, 0.1839, 0.1880, 0.0954, 0.1336, 0.0770], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0024, 0.0027, 0.0024, 0.0026, 0.0037, 0.0026, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:18:34,589 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9126, 2.8803, 2.1764, 2.9874, 2.8041, 2.8974, 2.6109, 2.2949], device='cuda:1'), covar=tensor([0.0925, 0.1287, 0.3000, 0.0667, 0.0924, 0.0831, 0.1230, 0.2528], device='cuda:1'), in_proj_covar=tensor([0.0264, 0.0294, 0.0266, 0.0245, 0.0303, 0.0288, 0.0251, 0.0255], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:19:13,452 INFO [train.py:873] (1/4) Epoch 11, batch 200, loss[loss=0.1394, simple_loss=0.1425, pruned_loss=0.06811, over 1252.00 frames. ], tot_loss[loss=0.1324, simple_loss=0.1608, pruned_loss=0.05195, over 1282138.61 frames. ], batch size: 100, lr: 7.37e-03, grad_scale: 8.0 2022-12-07 23:19:40,858 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 23:19:43,429 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.275e+02 2.829e+02 3.715e+02 6.696e+02, threshold=5.659e+02, percent-clipped=6.0 2022-12-07 23:20:05,291 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7532, 3.8150, 4.0304, 3.5453, 3.9194, 3.9427, 1.4407, 3.6895], device='cuda:1'), covar=tensor([0.0302, 0.0306, 0.0393, 0.0452, 0.0287, 0.0316, 0.3065, 0.0262], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0164, 0.0136, 0.0136, 0.0194, 0.0131, 0.0156, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:20:08,862 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:20:39,672 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0713, 2.3626, 2.4359, 2.5535, 2.0521, 2.5108, 2.1944, 1.2910], device='cuda:1'), covar=tensor([0.1424, 0.0890, 0.0880, 0.0524, 0.1126, 0.0561, 0.1526, 0.2534], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0075, 0.0059, 0.0063, 0.0090, 0.0072, 0.0093, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:1') 2022-12-07 23:20:41,220 INFO [train.py:873] (1/4) Epoch 11, batch 300, loss[loss=0.1357, simple_loss=0.1596, pruned_loss=0.05587, over 6908.00 frames. ], tot_loss[loss=0.1331, simple_loss=0.1613, pruned_loss=0.05252, over 1586337.84 frames. ], batch size: 100, lr: 7.37e-03, grad_scale: 8.0 2022-12-07 23:20:42,902 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 23:20:46,317 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:20:51,528 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:21:11,324 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.182e+02 2.664e+02 3.344e+02 7.501e+02, threshold=5.329e+02, percent-clipped=3.0 2022-12-07 23:21:28,768 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:22:02,491 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2568, 1.7264, 1.6932, 1.7611, 1.5712, 1.7764, 1.4451, 1.1399], device='cuda:1'), covar=tensor([0.1458, 0.1026, 0.0763, 0.0653, 0.1354, 0.0933, 0.2203, 0.2727], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0075, 0.0060, 0.0064, 0.0091, 0.0073, 0.0093, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-07 23:22:03,476 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6429, 1.5320, 1.6999, 1.4875, 1.8192, 0.9322, 1.5579, 1.5612], device='cuda:1'), covar=tensor([0.0837, 0.0917, 0.0679, 0.0868, 0.0776, 0.0870, 0.0881, 0.1231], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0027, 0.0024, 0.0025, 0.0037, 0.0026, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:22:10,197 INFO [train.py:873] (1/4) Epoch 11, batch 400, loss[loss=0.129, simple_loss=0.1616, pruned_loss=0.04824, over 6939.00 frames. ], tot_loss[loss=0.133, simple_loss=0.1609, pruned_loss=0.0526, over 1731346.67 frames. ], batch size: 100, lr: 7.36e-03, grad_scale: 8.0 2022-12-07 23:22:38,686 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4576, 4.4260, 5.1435, 5.3853, 5.1303, 4.5190, 5.2058, 4.5602], device='cuda:1'), covar=tensor([0.0679, 0.2136, 0.0729, 0.0916, 0.1127, 0.0610, 0.1304, 0.0926], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0260, 0.0185, 0.0177, 0.0177, 0.0145, 0.0268, 0.0160], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:22:40,270 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.357e+02 2.915e+02 3.546e+02 6.707e+02, threshold=5.830e+02, percent-clipped=5.0 2022-12-07 23:23:38,043 INFO [train.py:873] (1/4) Epoch 11, batch 500, loss[loss=0.1242, simple_loss=0.1352, pruned_loss=0.05663, over 2548.00 frames. ], tot_loss[loss=0.1323, simple_loss=0.1607, pruned_loss=0.05198, over 1854643.31 frames. ], batch size: 100, lr: 7.36e-03, grad_scale: 8.0 2022-12-07 23:23:58,784 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-07 23:24:01,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3543, 1.4791, 2.5362, 1.3561, 2.4681, 2.5219, 1.8647, 2.6178], device='cuda:1'), covar=tensor([0.0317, 0.2180, 0.0369, 0.1796, 0.0447, 0.0476, 0.1155, 0.0270], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0158, 0.0157, 0.0168, 0.0172, 0.0175, 0.0133, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:24:07,161 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2152, 2.9805, 2.7205, 2.8747, 3.1583, 3.1352, 3.2053, 3.1607], device='cuda:1'), covar=tensor([0.0882, 0.0733, 0.2269, 0.2640, 0.0816, 0.0942, 0.1058, 0.0882], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0247, 0.0419, 0.0539, 0.0312, 0.0401, 0.0384, 0.0346], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 23:24:07,886 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.432e+01 2.214e+02 2.601e+02 3.343e+02 5.170e+02, threshold=5.201e+02, percent-clipped=0.0 2022-12-07 23:24:10,887 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-07 23:24:47,799 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3624, 2.3625, 1.8898, 2.3952, 2.2380, 2.3573, 2.0937, 2.0266], device='cuda:1'), covar=tensor([0.0740, 0.0927, 0.2328, 0.0578, 0.0721, 0.0701, 0.1404, 0.1410], device='cuda:1'), in_proj_covar=tensor([0.0264, 0.0296, 0.0269, 0.0245, 0.0304, 0.0289, 0.0254, 0.0254], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:25:05,641 INFO [train.py:873] (1/4) Epoch 11, batch 600, loss[loss=0.1085, simple_loss=0.1484, pruned_loss=0.03429, over 14332.00 frames. ], tot_loss[loss=0.1324, simple_loss=0.1608, pruned_loss=0.05204, over 1884862.26 frames. ], batch size: 73, lr: 7.35e-03, grad_scale: 8.0 2022-12-07 23:25:35,309 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 2.199e+02 2.747e+02 3.310e+02 7.404e+02, threshold=5.494e+02, percent-clipped=4.0 2022-12-07 23:25:47,416 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:25:49,154 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 23:25:54,919 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-07 23:26:09,647 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:26:32,273 INFO [train.py:873] (1/4) Epoch 11, batch 700, loss[loss=0.1718, simple_loss=0.181, pruned_loss=0.08127, over 8620.00 frames. ], tot_loss[loss=0.1316, simple_loss=0.16, pruned_loss=0.05161, over 1889408.29 frames. ], batch size: 100, lr: 7.35e-03, grad_scale: 8.0 2022-12-07 23:26:34,466 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:26:40,819 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:26:43,434 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 23:27:01,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.063e+02 2.502e+02 2.955e+02 5.307e+02, threshold=5.004e+02, percent-clipped=0.0 2022-12-07 23:27:02,041 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:27:07,355 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2022-12-07 23:27:27,292 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:27:59,959 INFO [train.py:873] (1/4) Epoch 11, batch 800, loss[loss=0.1225, simple_loss=0.1496, pruned_loss=0.04767, over 13478.00 frames. ], tot_loss[loss=0.1326, simple_loss=0.1604, pruned_loss=0.05243, over 1910906.23 frames. ], batch size: 100, lr: 7.34e-03, grad_scale: 8.0 2022-12-07 23:28:08,579 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8350, 4.4188, 4.2892, 4.7746, 4.4802, 4.1747, 4.7752, 4.0218], device='cuda:1'), covar=tensor([0.0345, 0.0979, 0.0376, 0.0425, 0.0878, 0.0768, 0.0533, 0.0539], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0262, 0.0186, 0.0179, 0.0177, 0.0145, 0.0269, 0.0159], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:28:29,451 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.357e+02 2.806e+02 3.598e+02 6.382e+02, threshold=5.612e+02, percent-clipped=5.0 2022-12-07 23:28:33,934 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6714, 3.6742, 3.3231, 2.7043, 3.3272, 3.7049, 4.0669, 3.0141], device='cuda:1'), covar=tensor([0.0711, 0.1420, 0.0985, 0.1482, 0.0891, 0.0566, 0.0590, 0.1302], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0183, 0.0136, 0.0125, 0.0135, 0.0140, 0.0115, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0005, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 23:29:06,019 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6114, 2.3576, 2.4773, 1.6586, 2.2625, 2.5024, 2.6960, 2.1662], device='cuda:1'), covar=tensor([0.0710, 0.0907, 0.1044, 0.1817, 0.0950, 0.0635, 0.0608, 0.1426], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0185, 0.0136, 0.0127, 0.0136, 0.0141, 0.0116, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 23:29:26,690 INFO [train.py:873] (1/4) Epoch 11, batch 900, loss[loss=0.1187, simple_loss=0.1553, pruned_loss=0.04103, over 14281.00 frames. ], tot_loss[loss=0.1315, simple_loss=0.1598, pruned_loss=0.05155, over 1960627.81 frames. ], batch size: 80, lr: 7.34e-03, grad_scale: 8.0 2022-12-07 23:29:56,493 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.233e+02 2.846e+02 3.366e+02 5.399e+02, threshold=5.691e+02, percent-clipped=0.0 2022-12-07 23:30:54,355 INFO [train.py:873] (1/4) Epoch 11, batch 1000, loss[loss=0.1264, simple_loss=0.1607, pruned_loss=0.04606, over 14390.00 frames. ], tot_loss[loss=0.1319, simple_loss=0.1601, pruned_loss=0.05185, over 1909186.11 frames. ], batch size: 53, lr: 7.33e-03, grad_scale: 8.0 2022-12-07 23:30:57,558 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 23:30:57,834 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:31:19,965 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:31:24,209 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.175e+02 2.720e+02 3.257e+02 6.419e+02, threshold=5.440e+02, percent-clipped=2.0 2022-12-07 23:31:44,649 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:32:21,331 INFO [train.py:873] (1/4) Epoch 11, batch 1100, loss[loss=0.1828, simple_loss=0.1905, pruned_loss=0.08757, over 7765.00 frames. ], tot_loss[loss=0.1323, simple_loss=0.1603, pruned_loss=0.05221, over 1940700.58 frames. ], batch size: 100, lr: 7.33e-03, grad_scale: 8.0 2022-12-07 23:32:50,793 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.125e+02 2.272e+02 2.937e+02 3.615e+02 8.026e+02, threshold=5.874e+02, percent-clipped=3.0 2022-12-07 23:33:19,494 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1961, 1.2134, 1.4212, 0.9330, 0.9368, 1.2911, 0.7866, 1.1925], device='cuda:1'), covar=tensor([0.1383, 0.2637, 0.0926, 0.2290, 0.2602, 0.0747, 0.2150, 0.1107], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0095, 0.0092, 0.0096, 0.0114, 0.0082, 0.0128, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-07 23:33:44,644 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2090, 2.2074, 5.0128, 4.5307, 4.5594, 5.1363, 4.9550, 5.1812], device='cuda:1'), covar=tensor([0.1356, 0.1283, 0.0080, 0.0152, 0.0153, 0.0079, 0.0064, 0.0084], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0123, 0.0166, 0.0141, 0.0134, 0.0116, 0.0116], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:33:49,946 INFO [train.py:873] (1/4) Epoch 11, batch 1200, loss[loss=0.1678, simple_loss=0.1655, pruned_loss=0.08507, over 3862.00 frames. ], tot_loss[loss=0.1328, simple_loss=0.1606, pruned_loss=0.05252, over 1985419.48 frames. ], batch size: 100, lr: 7.32e-03, grad_scale: 8.0 2022-12-07 23:34:05,488 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:34:09,851 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2862, 2.5602, 2.4951, 2.6854, 1.9871, 2.7173, 2.3924, 1.0919], device='cuda:1'), covar=tensor([0.2408, 0.1040, 0.0996, 0.0722, 0.1499, 0.0658, 0.1758, 0.3390], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0075, 0.0060, 0.0063, 0.0091, 0.0073, 0.0094, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-07 23:34:19,397 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.391e+02 2.914e+02 3.659e+02 7.427e+02, threshold=5.827e+02, percent-clipped=2.0 2022-12-07 23:34:25,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2022-12-07 23:34:59,271 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:35:15,734 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3409, 3.4775, 3.6246, 3.2724, 3.5204, 3.3442, 1.3914, 3.3280], device='cuda:1'), covar=tensor([0.0323, 0.0335, 0.0405, 0.0530, 0.0321, 0.0526, 0.3168, 0.0303], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0164, 0.0137, 0.0136, 0.0193, 0.0131, 0.0156, 0.0181], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:35:17,347 INFO [train.py:873] (1/4) Epoch 11, batch 1300, loss[loss=0.1306, simple_loss=0.1606, pruned_loss=0.05024, over 14176.00 frames. ], tot_loss[loss=0.1319, simple_loss=0.1601, pruned_loss=0.05183, over 1985289.94 frames. ], batch size: 35, lr: 7.32e-03, grad_scale: 8.0 2022-12-07 23:35:20,642 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:35:42,636 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:35:47,102 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.104e+02 2.573e+02 3.185e+02 4.865e+02, threshold=5.146e+02, percent-clipped=0.0 2022-12-07 23:36:02,806 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:36:08,399 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:36:24,628 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:36:27,669 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3624, 1.9460, 3.5122, 2.3791, 3.3735, 1.8741, 2.6873, 3.3917], device='cuda:1'), covar=tensor([0.0743, 0.4472, 0.0505, 0.6130, 0.0659, 0.3636, 0.1288, 0.0573], device='cuda:1'), in_proj_covar=tensor([0.0242, 0.0215, 0.0197, 0.0289, 0.0216, 0.0216, 0.0212, 0.0204], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 23:36:45,316 INFO [train.py:873] (1/4) Epoch 11, batch 1400, loss[loss=0.1185, simple_loss=0.1565, pruned_loss=0.04027, over 14270.00 frames. ], tot_loss[loss=0.1323, simple_loss=0.1602, pruned_loss=0.0522, over 1973167.81 frames. ], batch size: 76, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:36:50,503 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:37:14,871 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.332e+02 2.864e+02 3.537e+02 8.782e+02, threshold=5.729e+02, percent-clipped=8.0 2022-12-07 23:37:49,621 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:37:57,474 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6373, 1.3864, 1.5900, 1.5402, 1.6115, 1.0070, 1.4879, 1.6750], device='cuda:1'), covar=tensor([0.0662, 0.1436, 0.0655, 0.0856, 0.0979, 0.0956, 0.0804, 0.0994], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0025, 0.0028, 0.0025, 0.0026, 0.0038, 0.0027, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:38:06,371 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0731, 4.7805, 4.6206, 5.0692, 4.6107, 4.2933, 5.0950, 4.9813], device='cuda:1'), covar=tensor([0.0606, 0.0688, 0.0709, 0.0570, 0.0862, 0.0490, 0.0562, 0.0706], device='cuda:1'), in_proj_covar=tensor([0.0130, 0.0125, 0.0133, 0.0142, 0.0135, 0.0110, 0.0156, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:38:12,197 INFO [train.py:873] (1/4) Epoch 11, batch 1500, loss[loss=0.141, simple_loss=0.161, pruned_loss=0.0605, over 8602.00 frames. ], tot_loss[loss=0.1328, simple_loss=0.1604, pruned_loss=0.05258, over 1937891.79 frames. ], batch size: 100, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:38:41,966 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.474e+02 2.983e+02 4.107e+02 8.785e+02, threshold=5.967e+02, percent-clipped=9.0 2022-12-07 23:38:43,077 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 23:38:43,519 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-07 23:38:51,452 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5433, 1.4015, 3.4605, 1.4414, 3.4045, 3.5920, 2.5811, 3.8466], device='cuda:1'), covar=tensor([0.0223, 0.3135, 0.0481, 0.2460, 0.0753, 0.0513, 0.0861, 0.0188], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0155, 0.0155, 0.0166, 0.0168, 0.0173, 0.0130, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:39:17,206 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:39:32,229 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:39:34,888 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0498, 1.6865, 1.6812, 1.8231, 1.6261, 1.7914, 2.2324, 2.0632], device='cuda:1'), covar=tensor([0.0532, 0.0799, 0.0727, 0.0495, 0.1174, 0.0561, 0.0346, 0.0647], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 23:39:40,746 INFO [train.py:873] (1/4) Epoch 11, batch 1600, loss[loss=0.1281, simple_loss=0.1394, pruned_loss=0.05842, over 2615.00 frames. ], tot_loss[loss=0.1319, simple_loss=0.1597, pruned_loss=0.05199, over 1942325.34 frames. ], batch size: 100, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:39:47,789 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9095, 3.0782, 4.7045, 3.5691, 4.7038, 4.5010, 4.4073, 4.0482], device='cuda:1'), covar=tensor([0.0581, 0.3224, 0.0961, 0.1814, 0.0774, 0.0849, 0.1883, 0.1827], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0321, 0.0399, 0.0306, 0.0381, 0.0322, 0.0364, 0.0315], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 23:39:53,208 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9979, 1.4441, 3.1139, 1.5252, 3.2227, 3.0923, 2.0468, 3.3084], device='cuda:1'), covar=tensor([0.0280, 0.2886, 0.0386, 0.2187, 0.0331, 0.0451, 0.1014, 0.0206], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0155, 0.0155, 0.0165, 0.0168, 0.0172, 0.0130, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:39:56,244 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-07 23:40:10,412 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.174e+02 1.982e+02 2.587e+02 3.263e+02 7.799e+02, threshold=5.175e+02, percent-clipped=3.0 2022-12-07 23:40:25,914 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8275, 1.5023, 1.7465, 1.6527, 1.3876, 0.9282, 1.5276, 1.8099], device='cuda:1'), covar=tensor([0.0817, 0.0929, 0.0474, 0.0876, 0.1765, 0.0803, 0.0976, 0.0581], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0025, 0.0027, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 23:40:25,930 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:26,944 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:30,495 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:44,742 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-07 23:41:08,994 INFO [train.py:873] (1/4) Epoch 11, batch 1700, loss[loss=0.1601, simple_loss=0.1428, pruned_loss=0.08868, over 1325.00 frames. ], tot_loss[loss=0.1308, simple_loss=0.1592, pruned_loss=0.05116, over 1918300.39 frames. ], batch size: 100, lr: 7.30e-03, grad_scale: 8.0 2022-12-07 23:41:12,498 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1914, 4.7368, 4.6030, 5.1361, 4.8424, 4.4188, 5.1248, 4.2095], device='cuda:1'), covar=tensor([0.0325, 0.0950, 0.0345, 0.0422, 0.0756, 0.0527, 0.0531, 0.0520], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0260, 0.0184, 0.0179, 0.0175, 0.0143, 0.0264, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:41:20,871 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:41:24,110 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:41:39,032 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.095e+02 2.641e+02 3.320e+02 6.995e+02, threshold=5.282e+02, percent-clipped=0.0 2022-12-07 23:41:53,245 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4856, 1.6075, 1.7869, 1.4322, 1.4644, 1.2531, 0.7550, 1.0693], device='cuda:1'), covar=tensor([0.0213, 0.0295, 0.0222, 0.0274, 0.0317, 0.0329, 0.0340, 0.0510], device='cuda:1'), in_proj_covar=tensor([0.0014, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 23:41:57,364 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4335, 1.1574, 2.0241, 1.8491, 1.8884, 2.0440, 1.2764, 2.0414], device='cuda:1'), covar=tensor([0.0696, 0.1211, 0.0197, 0.0390, 0.0449, 0.0186, 0.0687, 0.0244], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0125, 0.0169, 0.0142, 0.0136, 0.0117, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:42:37,339 INFO [train.py:873] (1/4) Epoch 11, batch 1800, loss[loss=0.1308, simple_loss=0.1625, pruned_loss=0.04953, over 14380.00 frames. ], tot_loss[loss=0.1321, simple_loss=0.1602, pruned_loss=0.052, over 1953469.76 frames. ], batch size: 73, lr: 7.30e-03, grad_scale: 8.0 2022-12-07 23:43:03,840 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.32 vs. limit=2.0 2022-12-07 23:43:04,162 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:43:07,257 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.136e+02 2.704e+02 3.283e+02 4.688e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-07 23:43:42,473 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:44:05,110 INFO [train.py:873] (1/4) Epoch 11, batch 1900, loss[loss=0.1564, simple_loss=0.1732, pruned_loss=0.0698, over 14139.00 frames. ], tot_loss[loss=0.1325, simple_loss=0.1604, pruned_loss=0.05228, over 1996363.68 frames. ], batch size: 99, lr: 7.29e-03, grad_scale: 8.0 2022-12-07 23:44:24,709 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:44:36,351 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.774e+01 2.327e+02 2.859e+02 3.630e+02 7.780e+02, threshold=5.718e+02, percent-clipped=3.0 2022-12-07 23:44:45,941 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:21,476 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8492, 1.5896, 1.7847, 2.0034, 1.4028, 1.6533, 1.8394, 1.9704], device='cuda:1'), covar=tensor([0.0106, 0.0240, 0.0101, 0.0095, 0.0208, 0.0277, 0.0125, 0.0117], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0251, 0.0370, 0.0317, 0.0260, 0.0301, 0.0292, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:45:25,658 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 23:45:29,366 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:33,420 INFO [train.py:873] (1/4) Epoch 11, batch 2000, loss[loss=0.1465, simple_loss=0.1705, pruned_loss=0.06127, over 14280.00 frames. ], tot_loss[loss=0.1328, simple_loss=0.1609, pruned_loss=0.05234, over 2030428.26 frames. ], batch size: 80, lr: 7.29e-03, grad_scale: 8.0 2022-12-07 23:45:40,499 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:44,062 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:46:04,192 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.333e+02 2.909e+02 3.438e+02 9.067e+02, threshold=5.818e+02, percent-clipped=5.0 2022-12-07 23:46:13,867 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1273, 1.8994, 2.1708, 2.3031, 1.8453, 1.8713, 2.1829, 2.1495], device='cuda:1'), covar=tensor([0.0206, 0.0429, 0.0189, 0.0186, 0.0333, 0.0597, 0.0252, 0.0259], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0250, 0.0369, 0.0315, 0.0258, 0.0299, 0.0290, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:46:16,077 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4939, 1.2353, 1.3973, 1.3773, 1.4226, 0.8893, 1.3313, 1.3084], device='cuda:1'), covar=tensor([0.0966, 0.1007, 0.0924, 0.0666, 0.0958, 0.1060, 0.1077, 0.1234], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0026, 0.0029, 0.0025, 0.0027, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:46:21,811 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:46:21,831 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:46:50,847 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6733, 4.2675, 4.0463, 4.6375, 4.3830, 4.1255, 4.6380, 3.8736], device='cuda:1'), covar=tensor([0.0371, 0.1024, 0.0437, 0.0479, 0.0810, 0.0691, 0.0626, 0.0560], device='cuda:1'), in_proj_covar=tensor([0.0168, 0.0259, 0.0184, 0.0181, 0.0175, 0.0144, 0.0266, 0.0159], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:46:55,432 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5006, 3.9203, 3.0622, 4.8052, 4.1740, 4.4287, 3.9025, 3.3413], device='cuda:1'), covar=tensor([0.0757, 0.1467, 0.3987, 0.0439, 0.1450, 0.1952, 0.1370, 0.3535], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0295, 0.0271, 0.0247, 0.0305, 0.0290, 0.0256, 0.0255], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-07 23:46:59,455 INFO [train.py:873] (1/4) Epoch 11, batch 2100, loss[loss=0.152, simple_loss=0.1744, pruned_loss=0.06481, over 9501.00 frames. ], tot_loss[loss=0.1326, simple_loss=0.1606, pruned_loss=0.05226, over 2024061.85 frames. ], batch size: 100, lr: 7.28e-03, grad_scale: 8.0 2022-12-07 23:47:14,322 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:47:25,615 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:47:29,642 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 2.408e+02 3.077e+02 4.325e+02 1.206e+03, threshold=6.153e+02, percent-clipped=10.0 2022-12-07 23:47:34,009 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2567, 1.8630, 2.1038, 1.3062, 1.9080, 2.1850, 2.3231, 1.8730], device='cuda:1'), covar=tensor([0.0643, 0.0819, 0.0887, 0.1756, 0.1060, 0.0660, 0.0469, 0.1419], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0179, 0.0132, 0.0123, 0.0131, 0.0137, 0.0114, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-07 23:48:07,065 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:48:26,366 INFO [train.py:873] (1/4) Epoch 11, batch 2200, loss[loss=0.1831, simple_loss=0.1901, pruned_loss=0.08804, over 8594.00 frames. ], tot_loss[loss=0.1337, simple_loss=0.1608, pruned_loss=0.05324, over 2000688.96 frames. ], batch size: 100, lr: 7.28e-03, grad_scale: 8.0 2022-12-07 23:48:57,072 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.100e+02 2.823e+02 3.472e+02 8.203e+02, threshold=5.646e+02, percent-clipped=4.0 2022-12-07 23:48:59,398 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2022-12-07 23:49:06,567 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:49:45,740 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8627, 1.7447, 4.4070, 4.1005, 4.0457, 4.5256, 4.0706, 4.4508], device='cuda:1'), covar=tensor([0.1650, 0.1658, 0.0132, 0.0234, 0.0221, 0.0128, 0.0168, 0.0162], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0125, 0.0168, 0.0143, 0.0136, 0.0119, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:49:48,090 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:49:53,346 INFO [train.py:873] (1/4) Epoch 11, batch 2300, loss[loss=0.1457, simple_loss=0.1549, pruned_loss=0.06827, over 5011.00 frames. ], tot_loss[loss=0.1327, simple_loss=0.1605, pruned_loss=0.05248, over 1990530.25 frames. ], batch size: 100, lr: 7.27e-03, grad_scale: 8.0 2022-12-07 23:50:00,421 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:04,152 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:05,082 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1587, 2.2148, 2.9321, 2.3678, 3.0404, 2.9668, 2.8123, 2.4810], device='cuda:1'), covar=tensor([0.0682, 0.2657, 0.1207, 0.1936, 0.0674, 0.0902, 0.1113, 0.1781], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0318, 0.0401, 0.0307, 0.0381, 0.0323, 0.0366, 0.0318], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-07 23:50:19,340 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 23:50:23,859 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.313e+02 2.906e+02 3.798e+02 8.346e+02, threshold=5.811e+02, percent-clipped=6.0 2022-12-07 23:50:38,065 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:42,321 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:45,017 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5537, 2.3010, 3.0631, 1.9538, 1.9098, 2.7638, 1.5341, 2.6657], device='cuda:1'), covar=tensor([0.1044, 0.1601, 0.0604, 0.2394, 0.2676, 0.0856, 0.4211, 0.0889], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0095, 0.0090, 0.0095, 0.0114, 0.0082, 0.0126, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-07 23:50:45,714 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:51,530 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 23:50:56,024 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 23:51:21,249 INFO [train.py:873] (1/4) Epoch 11, batch 2400, loss[loss=0.1006, simple_loss=0.1425, pruned_loss=0.02933, over 13971.00 frames. ], tot_loss[loss=0.1315, simple_loss=0.16, pruned_loss=0.05147, over 2021981.90 frames. ], batch size: 19, lr: 7.27e-03, grad_scale: 8.0 2022-12-07 23:51:30,815 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2022-12-07 23:51:32,083 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:51:38,900 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0504, 1.1409, 1.1731, 0.8828, 0.9567, 0.7188, 0.6440, 0.9068], device='cuda:1'), covar=tensor([0.0163, 0.0177, 0.0143, 0.0194, 0.0196, 0.0418, 0.0255, 0.0360], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 23:51:51,986 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.203e+02 2.586e+02 3.154e+02 7.524e+02, threshold=5.173e+02, percent-clipped=2.0 2022-12-07 23:52:14,423 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2022-12-07 23:52:49,055 INFO [train.py:873] (1/4) Epoch 11, batch 2500, loss[loss=0.1343, simple_loss=0.1635, pruned_loss=0.05258, over 14420.00 frames. ], tot_loss[loss=0.1312, simple_loss=0.1596, pruned_loss=0.05138, over 1930123.19 frames. ], batch size: 73, lr: 7.26e-03, grad_scale: 8.0 2022-12-07 23:53:03,260 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5193, 2.3413, 3.5735, 3.6763, 3.4441, 2.1977, 3.5929, 2.8378], device='cuda:1'), covar=tensor([0.0392, 0.0850, 0.0692, 0.0441, 0.0438, 0.1386, 0.0401, 0.0868], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0251, 0.0368, 0.0315, 0.0257, 0.0299, 0.0290, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:53:19,542 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.052e+02 2.774e+02 3.543e+02 6.822e+02, threshold=5.549e+02, percent-clipped=2.0 2022-12-07 23:54:02,588 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2095, 3.9508, 3.8341, 4.2637, 4.0510, 3.7909, 4.3034, 3.6063], device='cuda:1'), covar=tensor([0.0460, 0.0881, 0.0383, 0.0430, 0.0748, 0.1145, 0.0491, 0.0522], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0257, 0.0181, 0.0179, 0.0173, 0.0141, 0.0263, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:54:09,837 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5833, 4.2648, 4.1001, 4.6068, 4.2697, 4.0928, 4.6176, 3.9179], device='cuda:1'), covar=tensor([0.0371, 0.0910, 0.0390, 0.0421, 0.0841, 0.0748, 0.0513, 0.0492], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0258, 0.0181, 0.0180, 0.0174, 0.0141, 0.0264, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-07 23:54:16,599 INFO [train.py:873] (1/4) Epoch 11, batch 2600, loss[loss=0.1002, simple_loss=0.1426, pruned_loss=0.02891, over 13959.00 frames. ], tot_loss[loss=0.131, simple_loss=0.1597, pruned_loss=0.0511, over 1999835.16 frames. ], batch size: 26, lr: 7.26e-03, grad_scale: 8.0 2022-12-07 23:54:26,479 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:54:47,973 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.197e+02 2.874e+02 3.504e+02 7.274e+02, threshold=5.748e+02, percent-clipped=5.0 2022-12-07 23:54:51,605 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-07 23:54:59,990 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-07 23:55:00,960 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:55:19,415 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:55:41,776 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8397, 1.7336, 4.6391, 2.0072, 4.4541, 4.6139, 4.3993, 5.1299], device='cuda:1'), covar=tensor([0.0160, 0.2865, 0.0251, 0.2144, 0.0252, 0.0354, 0.0268, 0.0135], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0157, 0.0158, 0.0168, 0.0171, 0.0173, 0.0132, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:55:42,572 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:55:43,477 INFO [train.py:873] (1/4) Epoch 11, batch 2700, loss[loss=0.1087, simple_loss=0.1459, pruned_loss=0.03573, over 13967.00 frames. ], tot_loss[loss=0.1313, simple_loss=0.1595, pruned_loss=0.05153, over 1908465.91 frames. ], batch size: 20, lr: 7.25e-03, grad_scale: 4.0 2022-12-07 23:55:54,563 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:55:59,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0911, 1.1087, 1.0917, 0.8543, 0.9053, 0.7160, 0.8204, 0.8090], device='cuda:1'), covar=tensor([0.0186, 0.0183, 0.0175, 0.0163, 0.0183, 0.0419, 0.0258, 0.0335], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-07 23:56:15,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.898e+01 2.288e+02 2.843e+02 3.409e+02 7.759e+02, threshold=5.687e+02, percent-clipped=1.0 2022-12-07 23:56:28,046 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1606, 3.2738, 2.9928, 3.3537, 2.5929, 3.2606, 3.0922, 1.4608], device='cuda:1'), covar=tensor([0.1655, 0.0898, 0.1539, 0.0672, 0.1007, 0.0847, 0.1266, 0.2693], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0077, 0.0061, 0.0066, 0.0093, 0.0075, 0.0096, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-07 23:56:36,338 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:56:43,957 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0453, 1.8654, 1.6654, 1.5581, 1.8909, 1.0232, 1.8113, 1.8627], device='cuda:1'), covar=tensor([0.0774, 0.0770, 0.0715, 0.1364, 0.1425, 0.0891, 0.0978, 0.0909], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0025, 0.0027, 0.0024, 0.0026, 0.0037, 0.0025, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:57:11,629 INFO [train.py:873] (1/4) Epoch 11, batch 2800, loss[loss=0.1323, simple_loss=0.1411, pruned_loss=0.06177, over 2640.00 frames. ], tot_loss[loss=0.1317, simple_loss=0.1599, pruned_loss=0.05181, over 1908605.30 frames. ], batch size: 100, lr: 7.25e-03, grad_scale: 8.0 2022-12-07 23:57:43,401 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.311e+02 2.686e+02 3.123e+02 7.325e+02, threshold=5.372e+02, percent-clipped=4.0 2022-12-07 23:57:46,007 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:58:40,122 INFO [train.py:873] (1/4) Epoch 11, batch 2900, loss[loss=0.1453, simple_loss=0.1709, pruned_loss=0.05981, over 12791.00 frames. ], tot_loss[loss=0.1312, simple_loss=0.1593, pruned_loss=0.05158, over 1885173.74 frames. ], batch size: 100, lr: 7.24e-03, grad_scale: 8.0 2022-12-07 23:58:40,344 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:58:46,795 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5921, 4.3694, 4.2156, 4.6850, 4.3384, 3.7549, 4.6792, 4.4906], device='cuda:1'), covar=tensor([0.0673, 0.0726, 0.0747, 0.0566, 0.0676, 0.0727, 0.0577, 0.0780], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0126, 0.0134, 0.0144, 0.0134, 0.0111, 0.0157, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:58:54,229 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2701, 1.4102, 3.3174, 1.3412, 3.2156, 3.4218, 2.2664, 3.6057], device='cuda:1'), covar=tensor([0.0249, 0.3250, 0.0418, 0.2649, 0.0938, 0.0434, 0.0971, 0.0200], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0157, 0.0157, 0.0169, 0.0170, 0.0174, 0.0132, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-07 23:59:02,130 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2712, 2.3939, 4.1794, 4.3879, 4.2940, 2.5547, 4.3694, 3.2739], device='cuda:1'), covar=tensor([0.0306, 0.0966, 0.0774, 0.0340, 0.0308, 0.1414, 0.0278, 0.0800], device='cuda:1'), in_proj_covar=tensor([0.0280, 0.0249, 0.0367, 0.0314, 0.0256, 0.0297, 0.0290, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-07 23:59:04,624 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9215, 2.0269, 1.8392, 1.6786, 1.7491, 0.9585, 1.6235, 1.8911], device='cuda:1'), covar=tensor([0.1030, 0.0566, 0.0470, 0.1128, 0.2394, 0.0817, 0.1415, 0.1010], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0025, 0.0028, 0.0025, 0.0026, 0.0038, 0.0026, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-07 23:59:10,482 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:59:11,184 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.237e+02 2.916e+02 3.680e+02 6.575e+02, threshold=5.831e+02, percent-clipped=6.0 2022-12-07 23:59:15,163 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0458, 2.1828, 1.9639, 2.2168, 1.8877, 1.9823, 2.1383, 2.1359], device='cuda:1'), covar=tensor([0.0881, 0.0897, 0.1116, 0.0714, 0.1235, 0.0892, 0.0880, 0.0878], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0126, 0.0134, 0.0145, 0.0135, 0.0111, 0.0158, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-07 23:59:38,754 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:00:05,395 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:00:08,926 INFO [train.py:873] (1/4) Epoch 11, batch 3000, loss[loss=0.1016, simple_loss=0.1449, pruned_loss=0.02915, over 13994.00 frames. ], tot_loss[loss=0.1314, simple_loss=0.1595, pruned_loss=0.05168, over 1930007.75 frames. ], batch size: 22, lr: 7.24e-03, grad_scale: 8.0 2022-12-08 00:00:08,926 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 00:00:14,911 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0432, 2.0958, 1.9939, 1.7446, 2.2732, 1.1769, 2.0478, 2.0894], device='cuda:1'), covar=tensor([0.1164, 0.0679, 0.0576, 0.1516, 0.0598, 0.0768, 0.0806, 0.0551], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0025, 0.0027, 0.0025, 0.0026, 0.0038, 0.0026, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-08 00:00:15,267 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5789, 5.6169, 5.1079, 5.5576, 4.8588, 5.0085, 5.3847, 5.2755], device='cuda:1'), covar=tensor([0.0429, 0.0386, 0.0792, 0.0677, 0.0960, 0.0466, 0.0405, 0.0888], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0128, 0.0135, 0.0145, 0.0136, 0.0112, 0.0159, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 00:00:17,381 INFO [train.py:905] (1/4) Epoch 11, validation: loss=0.1282, simple_loss=0.1681, pruned_loss=0.04413, over 857387.00 frames. 2022-12-08 00:00:17,381 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 00:00:29,068 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-08 00:00:46,548 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 00:00:48,734 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.287e+02 2.909e+02 3.866e+02 8.652e+02, threshold=5.818e+02, percent-clipped=3.0 2022-12-08 00:01:46,386 INFO [train.py:873] (1/4) Epoch 11, batch 3100, loss[loss=0.131, simple_loss=0.1471, pruned_loss=0.05741, over 3897.00 frames. ], tot_loss[loss=0.1317, simple_loss=0.1599, pruned_loss=0.05175, over 1928117.88 frames. ], batch size: 100, lr: 7.24e-03, grad_scale: 8.0 2022-12-08 00:02:18,623 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.078e+02 2.524e+02 3.137e+02 7.110e+02, threshold=5.049e+02, percent-clipped=1.0 2022-12-08 00:02:45,178 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2022-12-08 00:02:57,658 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3976, 4.2265, 3.8213, 3.9375, 4.2359, 4.4435, 4.5455, 4.4593], device='cuda:1'), covar=tensor([0.1335, 0.0752, 0.2859, 0.3478, 0.1113, 0.1055, 0.1247, 0.1121], device='cuda:1'), in_proj_covar=tensor([0.0360, 0.0247, 0.0418, 0.0533, 0.0316, 0.0402, 0.0380, 0.0350], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:03:09,167 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 00:03:11,291 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:03:15,284 INFO [train.py:873] (1/4) Epoch 11, batch 3200, loss[loss=0.1354, simple_loss=0.1612, pruned_loss=0.0548, over 14171.00 frames. ], tot_loss[loss=0.1312, simple_loss=0.1595, pruned_loss=0.05142, over 1956770.13 frames. ], batch size: 99, lr: 7.23e-03, grad_scale: 8.0 2022-12-08 00:03:24,239 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0902, 1.9982, 4.1015, 2.7665, 3.9542, 1.9398, 3.0556, 3.9213], device='cuda:1'), covar=tensor([0.0631, 0.4765, 0.0550, 0.6736, 0.0635, 0.3836, 0.1273, 0.0448], device='cuda:1'), in_proj_covar=tensor([0.0247, 0.0219, 0.0201, 0.0293, 0.0221, 0.0218, 0.0214, 0.0204], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:03:46,996 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.001e+02 2.170e+02 2.756e+02 3.335e+02 6.522e+02, threshold=5.512e+02, percent-clipped=3.0 2022-12-08 00:03:55,368 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.94 vs. limit=5.0 2022-12-08 00:04:14,742 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:04:35,345 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:04:39,712 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 00:04:41,935 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9613, 1.7507, 4.7890, 2.5234, 4.3180, 4.8534, 4.6097, 5.3178], device='cuda:1'), covar=tensor([0.0186, 0.3188, 0.0353, 0.1907, 0.0370, 0.0451, 0.0282, 0.0183], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0157, 0.0158, 0.0167, 0.0171, 0.0174, 0.0131, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 00:04:43,622 INFO [train.py:873] (1/4) Epoch 11, batch 3300, loss[loss=0.1277, simple_loss=0.159, pruned_loss=0.0482, over 14173.00 frames. ], tot_loss[loss=0.1306, simple_loss=0.1592, pruned_loss=0.05099, over 1994056.99 frames. ], batch size: 89, lr: 7.23e-03, grad_scale: 8.0 2022-12-08 00:04:56,443 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:05:15,456 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.245e+02 2.789e+02 3.616e+02 5.775e+02, threshold=5.578e+02, percent-clipped=1.0 2022-12-08 00:05:16,283 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 00:05:22,618 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:05:47,002 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 00:06:07,618 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 00:06:10,656 INFO [train.py:873] (1/4) Epoch 11, batch 3400, loss[loss=0.0937, simple_loss=0.1384, pruned_loss=0.02449, over 13963.00 frames. ], tot_loss[loss=0.1302, simple_loss=0.1591, pruned_loss=0.05062, over 1987452.27 frames. ], batch size: 26, lr: 7.22e-03, grad_scale: 4.0 2022-12-08 00:06:15,632 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:06:34,641 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-08 00:06:43,352 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.208e+02 2.821e+02 3.585e+02 7.488e+02, threshold=5.641e+02, percent-clipped=3.0 2022-12-08 00:07:09,531 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7672, 2.6963, 2.5657, 2.8452, 2.3906, 2.4234, 2.8164, 2.7542], device='cuda:1'), covar=tensor([0.0799, 0.0980, 0.0975, 0.0736, 0.1207, 0.0955, 0.0829, 0.0871], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0127, 0.0134, 0.0145, 0.0135, 0.0111, 0.0157, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 00:07:24,696 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2022-12-08 00:07:34,152 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:07:38,504 INFO [train.py:873] (1/4) Epoch 11, batch 3500, loss[loss=0.1522, simple_loss=0.1624, pruned_loss=0.07101, over 6020.00 frames. ], tot_loss[loss=0.1282, simple_loss=0.1579, pruned_loss=0.04927, over 2009004.25 frames. ], batch size: 100, lr: 7.22e-03, grad_scale: 4.0 2022-12-08 00:07:53,818 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4817, 1.5921, 2.7617, 2.0825, 2.5664, 1.6704, 2.1111, 2.3983], device='cuda:1'), covar=tensor([0.1631, 0.4824, 0.0584, 0.5058, 0.1080, 0.4118, 0.1446, 0.0973], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0217, 0.0199, 0.0294, 0.0219, 0.0221, 0.0214, 0.0205], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:08:00,633 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.10 vs. limit=5.0 2022-12-08 00:08:06,214 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0647, 2.0591, 3.0611, 3.1571, 3.0652, 2.2206, 2.9987, 2.4836], device='cuda:1'), covar=tensor([0.0322, 0.0855, 0.0551, 0.0364, 0.0373, 0.1096, 0.0307, 0.0737], device='cuda:1'), in_proj_covar=tensor([0.0278, 0.0252, 0.0369, 0.0318, 0.0258, 0.0299, 0.0293, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:08:10,356 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.398e+02 2.883e+02 3.497e+02 6.738e+02, threshold=5.767e+02, percent-clipped=3.0 2022-12-08 00:08:15,410 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:08:57,429 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:08:58,201 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1491, 1.9773, 2.0604, 2.1704, 2.0611, 2.0512, 2.2179, 1.9227], device='cuda:1'), covar=tensor([0.0946, 0.1494, 0.0742, 0.0816, 0.1092, 0.0749, 0.0977, 0.0811], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0261, 0.0181, 0.0178, 0.0173, 0.0141, 0.0264, 0.0160], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 00:09:04,924 INFO [train.py:873] (1/4) Epoch 11, batch 3600, loss[loss=0.1816, simple_loss=0.189, pruned_loss=0.08708, over 11165.00 frames. ], tot_loss[loss=0.1291, simple_loss=0.1586, pruned_loss=0.0498, over 2002490.48 frames. ], batch size: 100, lr: 7.21e-03, grad_scale: 8.0 2022-12-08 00:09:05,124 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5127, 1.9967, 2.4984, 2.5369, 2.4586, 2.0360, 2.5721, 2.1500], device='cuda:1'), covar=tensor([0.0324, 0.0711, 0.0359, 0.0370, 0.0411, 0.0860, 0.0324, 0.0564], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0249, 0.0363, 0.0314, 0.0255, 0.0296, 0.0290, 0.0275], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:09:05,524 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 00:09:37,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.170e+02 2.534e+02 3.229e+02 7.657e+02, threshold=5.069e+02, percent-clipped=4.0 2022-12-08 00:09:39,212 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:09:47,258 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 00:10:01,648 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:10:32,868 INFO [train.py:873] (1/4) Epoch 11, batch 3700, loss[loss=0.148, simple_loss=0.1745, pruned_loss=0.06078, over 14354.00 frames. ], tot_loss[loss=0.1287, simple_loss=0.1586, pruned_loss=0.04938, over 2046574.31 frames. ], batch size: 73, lr: 7.21e-03, grad_scale: 8.0 2022-12-08 00:10:32,955 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:10:43,215 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:10:54,296 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:11:04,339 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.300e+02 2.713e+02 3.691e+02 7.855e+02, threshold=5.426e+02, percent-clipped=7.0 2022-12-08 00:11:09,815 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:11:19,281 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:11:35,773 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:11:58,465 INFO [train.py:873] (1/4) Epoch 11, batch 3800, loss[loss=0.156, simple_loss=0.1526, pruned_loss=0.07969, over 1230.00 frames. ], tot_loss[loss=0.1299, simple_loss=0.1593, pruned_loss=0.05021, over 2025163.03 frames. ], batch size: 100, lr: 7.20e-03, grad_scale: 8.0 2022-12-08 00:12:02,420 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:12:11,687 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:12:30,563 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.086e+02 2.258e+02 2.922e+02 3.679e+02 8.434e+02, threshold=5.844e+02, percent-clipped=6.0 2022-12-08 00:12:58,898 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-08 00:13:06,087 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8394, 1.5445, 3.7755, 3.4920, 3.6104, 3.7925, 3.1698, 3.8407], device='cuda:1'), covar=tensor([0.1458, 0.1460, 0.0118, 0.0221, 0.0216, 0.0125, 0.0246, 0.0110], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0127, 0.0166, 0.0143, 0.0135, 0.0117, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:13:26,570 INFO [train.py:873] (1/4) Epoch 11, batch 3900, loss[loss=0.108, simple_loss=0.15, pruned_loss=0.03299, over 13954.00 frames. ], tot_loss[loss=0.129, simple_loss=0.1583, pruned_loss=0.04984, over 2019959.67 frames. ], batch size: 26, lr: 7.20e-03, grad_scale: 8.0 2022-12-08 00:13:47,382 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2022-12-08 00:13:54,570 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 00:13:59,193 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.132e+02 2.623e+02 3.165e+02 6.507e+02, threshold=5.247e+02, percent-clipped=1.0 2022-12-08 00:14:22,690 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:14:54,982 INFO [train.py:873] (1/4) Epoch 11, batch 4000, loss[loss=0.09698, simple_loss=0.1393, pruned_loss=0.02735, over 13929.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1584, pruned_loss=0.0497, over 2056878.39 frames. ], batch size: 26, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:14:55,113 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:13,013 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:15:17,459 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:26,973 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:27,677 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.890e+01 2.190e+02 2.797e+02 3.644e+02 7.365e+02, threshold=5.595e+02, percent-clipped=3.0 2022-12-08 00:15:37,280 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:37,471 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3879, 2.6368, 4.1427, 3.1118, 4.2649, 4.0413, 3.9966, 3.6472], device='cuda:1'), covar=tensor([0.0714, 0.3616, 0.1110, 0.1886, 0.0809, 0.0928, 0.1459, 0.1932], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0318, 0.0397, 0.0306, 0.0379, 0.0321, 0.0363, 0.0312], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:15:55,378 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:21,115 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:23,021 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:23,755 INFO [train.py:873] (1/4) Epoch 11, batch 4100, loss[loss=0.1537, simple_loss=0.178, pruned_loss=0.06475, over 10317.00 frames. ], tot_loss[loss=0.1291, simple_loss=0.1586, pruned_loss=0.0498, over 2040010.90 frames. ], batch size: 100, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:16:32,687 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:16:56,454 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.218e+02 2.886e+02 3.615e+02 6.897e+02, threshold=5.771e+02, percent-clipped=4.0 2022-12-08 00:17:00,659 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:17:13,190 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 00:17:46,595 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9268, 1.5697, 2.1429, 1.7361, 1.9493, 1.4589, 1.7139, 1.9718], device='cuda:1'), covar=tensor([0.1670, 0.2341, 0.0333, 0.1492, 0.1041, 0.1187, 0.0954, 0.0759], device='cuda:1'), in_proj_covar=tensor([0.0244, 0.0215, 0.0196, 0.0291, 0.0217, 0.0217, 0.0214, 0.0201], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:17:52,398 INFO [train.py:873] (1/4) Epoch 11, batch 4200, loss[loss=0.1304, simple_loss=0.1568, pruned_loss=0.05205, over 12017.00 frames. ], tot_loss[loss=0.1296, simple_loss=0.159, pruned_loss=0.05009, over 1981706.62 frames. ], batch size: 100, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:17:54,342 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:18:08,601 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4002, 1.3760, 3.4788, 1.5019, 3.3433, 3.4819, 2.5165, 3.7809], device='cuda:1'), covar=tensor([0.0277, 0.3194, 0.0400, 0.2386, 0.0817, 0.0435, 0.0881, 0.0189], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0159, 0.0159, 0.0169, 0.0173, 0.0175, 0.0135, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 00:18:15,393 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2022-12-08 00:18:25,211 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.219e+02 2.651e+02 3.356e+02 1.038e+03, threshold=5.303e+02, percent-clipped=1.0 2022-12-08 00:18:55,927 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.29 vs. limit=2.0 2022-12-08 00:19:21,010 INFO [train.py:873] (1/4) Epoch 11, batch 4300, loss[loss=0.1312, simple_loss=0.1362, pruned_loss=0.06308, over 1257.00 frames. ], tot_loss[loss=0.1305, simple_loss=0.1591, pruned_loss=0.05093, over 1935011.78 frames. ], batch size: 100, lr: 7.18e-03, grad_scale: 8.0 2022-12-08 00:19:38,665 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:19:38,725 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:19:53,915 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.309e+02 2.745e+02 3.460e+02 6.233e+02, threshold=5.491e+02, percent-clipped=1.0 2022-12-08 00:20:21,811 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:20:21,853 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:38,364 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 00:20:47,239 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:53,266 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:53,975 INFO [train.py:873] (1/4) Epoch 11, batch 4400, loss[loss=0.1288, simple_loss=0.1627, pruned_loss=0.04746, over 14281.00 frames. ], tot_loss[loss=0.1302, simple_loss=0.1591, pruned_loss=0.05071, over 1935691.46 frames. ], batch size: 44, lr: 7.18e-03, grad_scale: 8.0 2022-12-08 00:20:54,072 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2738, 1.3300, 2.5222, 1.3325, 2.5143, 2.4931, 1.9992, 2.6158], device='cuda:1'), covar=tensor([0.0327, 0.2412, 0.0425, 0.1975, 0.0456, 0.0532, 0.1016, 0.0301], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0160, 0.0162, 0.0170, 0.0174, 0.0176, 0.0136, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 00:21:03,151 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:21:08,528 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:21:26,665 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.889e+01 2.388e+02 2.902e+02 3.759e+02 8.737e+02, threshold=5.804e+02, percent-clipped=8.0 2022-12-08 00:21:35,080 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:21:44,982 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:22:19,532 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:22:21,957 INFO [train.py:873] (1/4) Epoch 11, batch 4500, loss[loss=0.1181, simple_loss=0.1641, pruned_loss=0.03603, over 14388.00 frames. ], tot_loss[loss=0.1287, simple_loss=0.1582, pruned_loss=0.04958, over 1941873.65 frames. ], batch size: 53, lr: 7.17e-03, grad_scale: 8.0 2022-12-08 00:22:23,693 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.64 vs. limit=5.0 2022-12-08 00:22:35,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8436, 0.7710, 0.7625, 0.7669, 0.8392, 0.1989, 0.7455, 0.8173], device='cuda:1'), covar=tensor([0.0288, 0.0557, 0.0535, 0.0421, 0.0373, 0.0296, 0.0851, 0.0732], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0026, 0.0028, 0.0025, 0.0027, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-08 00:22:54,366 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.254e+02 2.921e+02 3.671e+02 7.258e+02, threshold=5.842e+02, percent-clipped=4.0 2022-12-08 00:22:54,567 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4704, 1.6736, 2.6987, 2.1515, 2.5097, 1.7926, 2.1836, 2.5068], device='cuda:1'), covar=tensor([0.1536, 0.3624, 0.0498, 0.3025, 0.0853, 0.2835, 0.0949, 0.0742], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0215, 0.0199, 0.0289, 0.0216, 0.0218, 0.0212, 0.0201], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:23:40,066 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1477, 2.1302, 2.3917, 1.4986, 1.5845, 2.1883, 1.2686, 2.1766], device='cuda:1'), covar=tensor([0.1136, 0.1515, 0.0877, 0.2535, 0.3054, 0.0921, 0.3990, 0.0946], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0094, 0.0091, 0.0094, 0.0114, 0.0083, 0.0124, 0.0087], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 00:23:48,489 INFO [train.py:873] (1/4) Epoch 11, batch 4600, loss[loss=0.1224, simple_loss=0.147, pruned_loss=0.04887, over 5953.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1584, pruned_loss=0.04969, over 1966001.20 frames. ], batch size: 100, lr: 7.17e-03, grad_scale: 8.0 2022-12-08 00:24:06,631 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:24:20,612 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2008, 4.1847, 4.5167, 3.8444, 4.3078, 4.5420, 1.7005, 3.9836], device='cuda:1'), covar=tensor([0.0264, 0.0325, 0.0339, 0.0471, 0.0306, 0.0220, 0.2988, 0.0281], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0167, 0.0139, 0.0137, 0.0197, 0.0132, 0.0156, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:24:21,385 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.172e+02 2.626e+02 3.163e+02 7.453e+02, threshold=5.252e+02, percent-clipped=2.0 2022-12-08 00:24:42,626 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:24:48,269 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:08,282 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8754, 0.8006, 0.7343, 0.8576, 0.8095, 0.1527, 0.7114, 0.8011], device='cuda:1'), covar=tensor([0.0280, 0.0460, 0.0458, 0.0326, 0.0215, 0.0232, 0.0746, 0.0597], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0025, 0.0028, 0.0025, 0.0026, 0.0038, 0.0026, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:1') 2022-12-08 00:25:09,123 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:15,102 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 00:25:16,060 INFO [train.py:873] (1/4) Epoch 11, batch 4700, loss[loss=0.1449, simple_loss=0.1473, pruned_loss=0.0713, over 2680.00 frames. ], tot_loss[loss=0.1301, simple_loss=0.1588, pruned_loss=0.05069, over 1917159.46 frames. ], batch size: 100, lr: 7.16e-03, grad_scale: 8.0 2022-12-08 00:25:36,798 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:49,171 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 2.137e+02 2.805e+02 3.644e+02 8.749e+02, threshold=5.610e+02, percent-clipped=7.0 2022-12-08 00:25:51,854 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:03,741 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8056, 1.7581, 3.0550, 2.2836, 2.8325, 1.8572, 2.4143, 2.8480], device='cuda:1'), covar=tensor([0.1033, 0.4100, 0.0631, 0.4452, 0.0737, 0.3066, 0.1212, 0.0684], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0216, 0.0199, 0.0290, 0.0216, 0.0218, 0.0212, 0.0203], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:26:20,343 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:42,728 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:45,223 INFO [train.py:873] (1/4) Epoch 11, batch 4800, loss[loss=0.1314, simple_loss=0.1462, pruned_loss=0.05828, over 3901.00 frames. ], tot_loss[loss=0.1318, simple_loss=0.1596, pruned_loss=0.05198, over 1881090.09 frames. ], batch size: 100, lr: 7.16e-03, grad_scale: 8.0 2022-12-08 00:27:14,134 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:17,205 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.171e+02 2.910e+02 3.468e+02 6.443e+02, threshold=5.821e+02, percent-clipped=1.0 2022-12-08 00:27:21,566 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:23,922 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:45,721 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-08 00:27:49,327 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:01,205 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3383, 1.8341, 2.3115, 2.0222, 2.3685, 2.2458, 2.1683, 2.1013], device='cuda:1'), covar=tensor([0.0482, 0.2255, 0.0661, 0.1426, 0.0488, 0.0846, 0.0519, 0.1324], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0313, 0.0397, 0.0304, 0.0375, 0.0318, 0.0360, 0.0313], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:28:06,895 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6169, 5.3092, 4.9829, 5.2217, 5.1218, 5.4580, 5.5332, 5.6139], device='cuda:1'), covar=tensor([0.0612, 0.0483, 0.2007, 0.2470, 0.0697, 0.0610, 0.1045, 0.0705], device='cuda:1'), in_proj_covar=tensor([0.0366, 0.0252, 0.0429, 0.0539, 0.0327, 0.0411, 0.0390, 0.0354], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:28:12,374 INFO [train.py:873] (1/4) Epoch 11, batch 4900, loss[loss=0.1168, simple_loss=0.1524, pruned_loss=0.04059, over 14355.00 frames. ], tot_loss[loss=0.1313, simple_loss=0.1598, pruned_loss=0.05143, over 1932722.80 frames. ], batch size: 28, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:28:14,956 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:37,202 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8600, 1.7286, 1.8828, 1.7537, 2.0329, 1.8563, 1.7167, 1.8666], device='cuda:1'), covar=tensor([0.0523, 0.1268, 0.0278, 0.0466, 0.0360, 0.0660, 0.0317, 0.0401], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0314, 0.0400, 0.0306, 0.0377, 0.0320, 0.0362, 0.0313], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:28:42,475 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:44,856 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.165e+02 2.757e+02 3.267e+02 6.600e+02, threshold=5.514e+02, percent-clipped=1.0 2022-12-08 00:28:53,552 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.57 vs. limit=5.0 2022-12-08 00:29:02,799 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9561, 3.7524, 3.7291, 4.0542, 3.7100, 3.3716, 4.0503, 3.8965], device='cuda:1'), covar=tensor([0.0719, 0.0906, 0.0794, 0.0613, 0.0789, 0.0799, 0.0660, 0.0747], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0127, 0.0132, 0.0143, 0.0134, 0.0112, 0.0156, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 00:29:41,614 INFO [train.py:873] (1/4) Epoch 11, batch 5000, loss[loss=0.162, simple_loss=0.1721, pruned_loss=0.07597, over 7742.00 frames. ], tot_loss[loss=0.131, simple_loss=0.1591, pruned_loss=0.05143, over 1866469.12 frames. ], batch size: 100, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:29:57,590 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:30:14,652 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.220e+02 2.746e+02 3.480e+02 5.243e+02, threshold=5.491e+02, percent-clipped=0.0 2022-12-08 00:31:10,693 INFO [train.py:873] (1/4) Epoch 11, batch 5100, loss[loss=0.1278, simple_loss=0.1615, pruned_loss=0.04704, over 14278.00 frames. ], tot_loss[loss=0.1304, simple_loss=0.1587, pruned_loss=0.05103, over 1901189.08 frames. ], batch size: 31, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:31:24,289 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:31:35,018 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:31:43,012 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.114e+02 2.657e+02 3.455e+02 5.035e+02, threshold=5.315e+02, percent-clipped=0.0 2022-12-08 00:31:56,800 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:17,382 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:36,968 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:38,683 INFO [train.py:873] (1/4) Epoch 11, batch 5200, loss[loss=0.1396, simple_loss=0.1642, pruned_loss=0.05751, over 14220.00 frames. ], tot_loss[loss=0.1296, simple_loss=0.1587, pruned_loss=0.05024, over 1998935.58 frames. ], batch size: 89, lr: 7.14e-03, grad_scale: 8.0 2022-12-08 00:32:45,324 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 00:32:50,379 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:33:04,574 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:33:05,087 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 00:33:11,238 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.222e+02 2.775e+02 3.597e+02 5.956e+02, threshold=5.550e+02, percent-clipped=3.0 2022-12-08 00:33:54,522 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6016, 1.2377, 2.0789, 1.8538, 1.9386, 2.0825, 1.4182, 2.1055], device='cuda:1'), covar=tensor([0.0695, 0.1030, 0.0203, 0.0379, 0.0398, 0.0222, 0.0645, 0.0209], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0125, 0.0166, 0.0141, 0.0135, 0.0117, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:34:04,553 INFO [train.py:873] (1/4) Epoch 11, batch 5300, loss[loss=0.109, simple_loss=0.1458, pruned_loss=0.03605, over 14250.00 frames. ], tot_loss[loss=0.1283, simple_loss=0.1581, pruned_loss=0.04927, over 2027130.26 frames. ], batch size: 39, lr: 7.14e-03, grad_scale: 8.0 2022-12-08 00:34:07,202 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.41 vs. limit=2.0 2022-12-08 00:34:20,713 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:34:35,533 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5042, 1.6562, 4.3527, 2.2210, 4.2236, 4.4261, 4.0126, 4.8716], device='cuda:1'), covar=tensor([0.0201, 0.3290, 0.0337, 0.2125, 0.0300, 0.0348, 0.0358, 0.0135], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0157, 0.0167, 0.0169, 0.0173, 0.0134, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 00:34:37,142 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.266e+02 2.683e+02 3.445e+02 6.231e+02, threshold=5.367e+02, percent-clipped=3.0 2022-12-08 00:35:02,411 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:35:28,111 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2022-12-08 00:35:32,523 INFO [train.py:873] (1/4) Epoch 11, batch 5400, loss[loss=0.1331, simple_loss=0.15, pruned_loss=0.05812, over 5975.00 frames. ], tot_loss[loss=0.1283, simple_loss=0.1582, pruned_loss=0.04915, over 2054299.07 frames. ], batch size: 100, lr: 7.13e-03, grad_scale: 16.0 2022-12-08 00:35:58,281 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81049.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:03,820 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-08 00:36:05,567 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.933e+01 1.955e+02 2.740e+02 3.056e+02 5.675e+02, threshold=5.479e+02, percent-clipped=1.0 2022-12-08 00:36:26,090 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2022-12-08 00:36:35,743 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:39,863 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:56,391 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:58,867 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81119.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:00,419 INFO [train.py:873] (1/4) Epoch 11, batch 5500, loss[loss=0.1454, simple_loss=0.1681, pruned_loss=0.06139, over 9540.00 frames. ], tot_loss[loss=0.1278, simple_loss=0.1577, pruned_loss=0.04896, over 2026207.28 frames. ], batch size: 100, lr: 7.13e-03, grad_scale: 8.0 2022-12-08 00:37:02,700 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2638, 2.0810, 4.6837, 4.4483, 4.2821, 4.8288, 4.4948, 4.8445], device='cuda:1'), covar=tensor([0.1334, 0.1296, 0.0087, 0.0168, 0.0171, 0.0082, 0.0140, 0.0089], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0125, 0.0165, 0.0141, 0.0135, 0.0118, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:37:07,721 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81129.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:14,079 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 00:37:25,926 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81150.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:33,796 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.115e+02 2.510e+02 3.295e+02 5.828e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 00:37:40,964 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81167.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:49,664 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81177.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:38:07,658 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81198.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:38:28,043 INFO [train.py:873] (1/4) Epoch 11, batch 5600, loss[loss=0.1156, simple_loss=0.1516, pruned_loss=0.03987, over 14164.00 frames. ], tot_loss[loss=0.1284, simple_loss=0.1577, pruned_loss=0.04954, over 2004194.44 frames. ], batch size: 99, lr: 7.12e-03, grad_scale: 8.0 2022-12-08 00:39:02,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.253e+02 2.769e+02 3.640e+02 8.621e+02, threshold=5.537e+02, percent-clipped=7.0 2022-12-08 00:39:24,861 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 00:39:37,276 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:39:38,944 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81301.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:39:55,988 INFO [train.py:873] (1/4) Epoch 11, batch 5700, loss[loss=0.1229, simple_loss=0.1481, pruned_loss=0.04882, over 6920.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1582, pruned_loss=0.05006, over 2052570.39 frames. ], batch size: 100, lr: 7.12e-03, grad_scale: 8.0 2022-12-08 00:40:07,939 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0500, 1.9081, 4.7917, 4.4601, 4.3956, 4.9133, 4.6958, 4.8827], device='cuda:1'), covar=tensor([0.1428, 0.1412, 0.0096, 0.0153, 0.0161, 0.0098, 0.0083, 0.0123], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0158, 0.0124, 0.0165, 0.0140, 0.0134, 0.0117, 0.0117], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 00:40:24,078 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-08 00:40:28,773 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.149e+01 2.189e+02 2.731e+02 3.467e+02 4.923e+02, threshold=5.462e+02, percent-clipped=0.0 2022-12-08 00:40:29,816 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81360.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:40:31,310 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81362.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:40:53,440 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0643, 4.8315, 4.5962, 5.0618, 4.7012, 4.3871, 5.1589, 4.9739], device='cuda:1'), covar=tensor([0.0565, 0.0609, 0.0783, 0.0568, 0.0801, 0.0540, 0.0479, 0.0584], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0127, 0.0135, 0.0144, 0.0135, 0.0112, 0.0154, 0.0132], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 00:40:57,935 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81392.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:23,788 INFO [train.py:873] (1/4) Epoch 11, batch 5800, loss[loss=0.1366, simple_loss=0.1614, pruned_loss=0.05591, over 6906.00 frames. ], tot_loss[loss=0.13, simple_loss=0.159, pruned_loss=0.05057, over 2015294.56 frames. ], batch size: 100, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:41:30,636 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:40,074 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:40,959 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6342, 3.3764, 3.3019, 3.6258, 3.4392, 3.6224, 3.6706, 3.1094], device='cuda:1'), covar=tensor([0.0504, 0.1027, 0.0513, 0.0540, 0.0803, 0.0388, 0.0643, 0.0573], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0256, 0.0180, 0.0178, 0.0173, 0.0143, 0.0262, 0.0157], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 00:41:56,860 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.147e+02 2.711e+02 3.541e+02 7.030e+02, threshold=5.421e+02, percent-clipped=5.0 2022-12-08 00:42:08,748 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:42:12,683 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81477.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:42:51,270 INFO [train.py:873] (1/4) Epoch 11, batch 5900, loss[loss=0.1416, simple_loss=0.1672, pruned_loss=0.058, over 11231.00 frames. ], tot_loss[loss=0.13, simple_loss=0.159, pruned_loss=0.05053, over 2002457.69 frames. ], batch size: 100, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:43:03,450 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2022-12-08 00:43:24,407 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.246e+02 2.674e+02 3.531e+02 7.780e+02, threshold=5.349e+02, percent-clipped=2.0 2022-12-08 00:44:19,735 INFO [train.py:873] (1/4) Epoch 11, batch 6000, loss[loss=0.1082, simple_loss=0.1201, pruned_loss=0.04818, over 2650.00 frames. ], tot_loss[loss=0.1301, simple_loss=0.1586, pruned_loss=0.05077, over 1940276.44 frames. ], batch size: 100, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:44:19,735 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 00:44:28,212 INFO [train.py:905] (1/4) Epoch 11, validation: loss=0.1286, simple_loss=0.169, pruned_loss=0.04409, over 857387.00 frames. 2022-12-08 00:44:28,213 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 00:44:39,548 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5799, 1.4810, 1.6825, 1.4888, 1.6413, 1.0138, 1.4170, 1.5959], device='cuda:1'), covar=tensor([0.1282, 0.1077, 0.0886, 0.0881, 0.0785, 0.0896, 0.1305, 0.0683], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 00:44:58,298 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81655.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:00,143 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81657.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:01,818 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.152e+02 2.716e+02 3.284e+02 6.077e+02, threshold=5.431e+02, percent-clipped=1.0 2022-12-08 00:45:11,815 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81670.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:21,338 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6698, 3.0055, 4.4365, 3.4004, 4.4210, 4.3735, 4.2080, 3.8155], device='cuda:1'), covar=tensor([0.0695, 0.2693, 0.0971, 0.1667, 0.0819, 0.0762, 0.1940, 0.1713], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0314, 0.0399, 0.0305, 0.0376, 0.0321, 0.0366, 0.0311], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:45:30,043 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8451, 3.1356, 4.4861, 3.5212, 4.5542, 4.6459, 4.4047, 4.0005], device='cuda:1'), covar=tensor([0.0595, 0.2849, 0.1220, 0.1598, 0.0781, 0.0712, 0.1491, 0.1688], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0314, 0.0399, 0.0305, 0.0376, 0.0320, 0.0366, 0.0311], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:45:34,443 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3147, 1.3027, 1.2848, 1.3161, 1.3355, 0.8136, 1.1171, 1.0959], device='cuda:1'), covar=tensor([0.0584, 0.0713, 0.0599, 0.0665, 0.0492, 0.0930, 0.0941, 0.0946], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0027, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-08 00:45:55,906 INFO [train.py:873] (1/4) Epoch 11, batch 6100, loss[loss=0.1671, simple_loss=0.1567, pruned_loss=0.08874, over 1236.00 frames. ], tot_loss[loss=0.1291, simple_loss=0.158, pruned_loss=0.05011, over 1944886.29 frames. ], batch size: 100, lr: 7.10e-03, grad_scale: 8.0 2022-12-08 00:46:04,771 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81731.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:46:29,253 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.193e+02 2.528e+02 3.178e+02 6.202e+02, threshold=5.055e+02, percent-clipped=2.0 2022-12-08 00:46:35,148 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2022-12-08 00:46:40,596 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:47:09,959 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2022-12-08 00:47:13,013 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0300, 1.7852, 1.8953, 1.6795, 2.0472, 1.1799, 1.8009, 1.8321], device='cuda:1'), covar=tensor([0.1115, 0.1004, 0.0636, 0.3242, 0.0826, 0.0835, 0.1106, 0.0709], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0028, 0.0025, 0.0027, 0.0038, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-08 00:47:23,447 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:47:24,228 INFO [train.py:873] (1/4) Epoch 11, batch 6200, loss[loss=0.1281, simple_loss=0.1635, pruned_loss=0.04639, over 14413.00 frames. ], tot_loss[loss=0.1307, simple_loss=0.1591, pruned_loss=0.05119, over 1918267.06 frames. ], batch size: 41, lr: 7.10e-03, grad_scale: 8.0 2022-12-08 00:47:58,181 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.324e+02 2.708e+02 3.471e+02 1.552e+03, threshold=5.417e+02, percent-clipped=7.0 2022-12-08 00:48:03,815 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 00:48:13,443 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2913, 1.6716, 2.5594, 2.1022, 2.2981, 1.6942, 1.9944, 2.3145], device='cuda:1'), covar=tensor([0.1801, 0.3106, 0.0514, 0.2361, 0.0913, 0.2003, 0.1114, 0.0735], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0213, 0.0201, 0.0290, 0.0217, 0.0216, 0.0216, 0.0203], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:48:20,270 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8299, 1.1896, 1.3527, 1.3223, 1.0105, 1.3237, 1.1028, 0.8161], device='cuda:1'), covar=tensor([0.1982, 0.0933, 0.0417, 0.0393, 0.1524, 0.0628, 0.1591, 0.1503], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0077, 0.0061, 0.0065, 0.0092, 0.0075, 0.0094, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 00:48:33,521 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-08 00:48:43,711 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.58 vs. limit=5.0 2022-12-08 00:48:52,507 INFO [train.py:873] (1/4) Epoch 11, batch 6300, loss[loss=0.1487, simple_loss=0.164, pruned_loss=0.06669, over 4929.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1581, pruned_loss=0.05014, over 1914163.81 frames. ], batch size: 100, lr: 7.09e-03, grad_scale: 8.0 2022-12-08 00:48:58,879 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2022-12-08 00:49:05,724 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6736, 4.4391, 4.1180, 4.3092, 4.4579, 4.5738, 4.7068, 4.6343], device='cuda:1'), covar=tensor([0.0822, 0.0499, 0.2033, 0.2410, 0.0644, 0.0701, 0.0694, 0.0754], device='cuda:1'), in_proj_covar=tensor([0.0370, 0.0260, 0.0430, 0.0544, 0.0323, 0.0417, 0.0389, 0.0362], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:49:22,507 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81955.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:49:24,266 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81957.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:49:25,808 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.360e+02 2.936e+02 3.429e+02 7.128e+02, threshold=5.872e+02, percent-clipped=5.0 2022-12-08 00:49:50,570 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:04,809 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:06,454 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82005.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:20,250 INFO [train.py:873] (1/4) Epoch 11, batch 6400, loss[loss=0.1049, simple_loss=0.1422, pruned_loss=0.03376, over 13883.00 frames. ], tot_loss[loss=0.1282, simple_loss=0.1578, pruned_loss=0.04934, over 1938068.49 frames. ], batch size: 23, lr: 7.09e-03, grad_scale: 8.0 2022-12-08 00:50:24,708 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82026.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 00:50:43,929 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82048.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:50,079 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:53,436 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.188e+02 2.645e+02 3.155e+02 5.407e+02, threshold=5.290e+02, percent-clipped=0.0 2022-12-08 00:51:19,294 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 00:51:43,602 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:51:47,737 INFO [train.py:873] (1/4) Epoch 11, batch 6500, loss[loss=0.1211, simple_loss=0.156, pruned_loss=0.04306, over 14279.00 frames. ], tot_loss[loss=0.1281, simple_loss=0.1578, pruned_loss=0.04915, over 1954902.72 frames. ], batch size: 31, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:51:57,510 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 00:52:07,298 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0155, 3.9952, 4.3333, 3.5987, 4.1097, 4.3280, 1.6482, 3.8865], device='cuda:1'), covar=tensor([0.0308, 0.0466, 0.0340, 0.0515, 0.0361, 0.0270, 0.3061, 0.0301], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0168, 0.0139, 0.0136, 0.0199, 0.0133, 0.0158, 0.0183], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 00:52:15,472 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8267, 0.7915, 0.7347, 0.8441, 0.7564, 0.1703, 0.6694, 0.8484], device='cuda:1'), covar=tensor([0.0267, 0.0522, 0.0494, 0.0417, 0.0234, 0.0288, 0.0800, 0.0572], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 00:52:20,599 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.664e+01 2.301e+02 3.022e+02 3.856e+02 8.544e+02, threshold=6.044e+02, percent-clipped=9.0 2022-12-08 00:52:55,927 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8547, 3.5389, 2.7900, 4.0615, 3.9372, 3.8804, 3.3578, 2.7288], device='cuda:1'), covar=tensor([0.0948, 0.1473, 0.3842, 0.0633, 0.0931, 0.1345, 0.1307, 0.3602], device='cuda:1'), in_proj_covar=tensor([0.0268, 0.0298, 0.0271, 0.0257, 0.0314, 0.0300, 0.0254, 0.0255], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-08 00:53:02,167 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7922, 1.5626, 1.8987, 1.7982, 1.7189, 1.6084, 1.4927, 1.0821], device='cuda:1'), covar=tensor([0.0259, 0.0634, 0.0366, 0.0566, 0.0340, 0.0263, 0.0263, 0.0665], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0015, 0.0015, 0.0025, 0.0020, 0.0025], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:1') 2022-12-08 00:53:14,952 INFO [train.py:873] (1/4) Epoch 11, batch 6600, loss[loss=0.1014, simple_loss=0.1438, pruned_loss=0.02949, over 14350.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.158, pruned_loss=0.05015, over 1897801.79 frames. ], batch size: 31, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:53:20,506 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.40 vs. limit=5.0 2022-12-08 00:53:48,293 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.109e+02 2.550e+02 3.112e+02 5.386e+02, threshold=5.100e+02, percent-clipped=0.0 2022-12-08 00:53:48,490 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9655, 1.6695, 1.9648, 1.6876, 2.0173, 1.8801, 1.7203, 1.9631], device='cuda:1'), covar=tensor([0.0599, 0.1424, 0.0280, 0.0432, 0.0386, 0.0605, 0.0287, 0.0275], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0314, 0.0399, 0.0309, 0.0377, 0.0322, 0.0367, 0.0312], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:54:09,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7883, 3.5319, 3.4436, 3.7590, 3.6035, 3.8077, 3.7948, 3.1813], device='cuda:1'), covar=tensor([0.0428, 0.0974, 0.0432, 0.0490, 0.0690, 0.0273, 0.0634, 0.0555], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0253, 0.0177, 0.0177, 0.0169, 0.0142, 0.0262, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 00:54:43,678 INFO [train.py:873] (1/4) Epoch 11, batch 6700, loss[loss=0.1227, simple_loss=0.1562, pruned_loss=0.04463, over 14183.00 frames. ], tot_loss[loss=0.1295, simple_loss=0.1585, pruned_loss=0.05021, over 1951427.60 frames. ], batch size: 99, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:54:48,064 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82326.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:02,411 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82343.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:05,072 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:16,416 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.173e+02 2.670e+02 3.578e+02 7.367e+02, threshold=5.339e+02, percent-clipped=2.0 2022-12-08 00:55:29,455 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:59,200 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:56:02,574 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82411.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:56:11,098 INFO [train.py:873] (1/4) Epoch 11, batch 6800, loss[loss=0.1329, simple_loss=0.1647, pruned_loss=0.05061, over 14526.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.158, pruned_loss=0.05019, over 1895074.00 frames. ], batch size: 51, lr: 7.07e-03, grad_scale: 8.0 2022-12-08 00:56:13,236 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0450, 2.1202, 2.8471, 2.3033, 2.8637, 2.8222, 2.7003, 2.4380], device='cuda:1'), covar=tensor([0.0754, 0.2666, 0.0900, 0.1763, 0.0774, 0.0977, 0.1169, 0.1776], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0316, 0.0402, 0.0311, 0.0379, 0.0326, 0.0370, 0.0315], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 00:56:43,724 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.471e+02 3.089e+02 4.263e+02 8.393e+02, threshold=6.178e+02, percent-clipped=10.0 2022-12-08 00:57:02,153 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 00:57:14,081 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82493.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:57:25,018 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7317, 1.5646, 1.6268, 1.6414, 1.6544, 0.9961, 1.7008, 1.7995], device='cuda:1'), covar=tensor([0.0797, 0.1153, 0.1268, 0.1132, 0.1414, 0.0949, 0.0668, 0.1031], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 00:57:35,449 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8655, 2.6967, 2.7543, 2.9277, 2.7969, 2.8576, 2.9699, 2.4776], device='cuda:1'), covar=tensor([0.0672, 0.1152, 0.0587, 0.0593, 0.0800, 0.0557, 0.0678, 0.0652], device='cuda:1'), in_proj_covar=tensor([0.0167, 0.0258, 0.0181, 0.0181, 0.0173, 0.0145, 0.0268, 0.0160], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 00:57:37,891 INFO [train.py:873] (1/4) Epoch 11, batch 6900, loss[loss=0.1192, simple_loss=0.1483, pruned_loss=0.045, over 14298.00 frames. ], tot_loss[loss=0.13, simple_loss=0.1583, pruned_loss=0.05082, over 1962891.69 frames. ], batch size: 66, lr: 7.07e-03, grad_scale: 8.0 2022-12-08 00:58:06,352 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82554.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:58:11,135 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.298e+02 3.049e+02 3.803e+02 8.603e+02, threshold=6.098e+02, percent-clipped=2.0 2022-12-08 00:58:20,119 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82569.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:59:06,536 INFO [train.py:873] (1/4) Epoch 11, batch 7000, loss[loss=0.1547, simple_loss=0.1466, pruned_loss=0.08143, over 2618.00 frames. ], tot_loss[loss=0.1294, simple_loss=0.158, pruned_loss=0.05037, over 1952325.82 frames. ], batch size: 100, lr: 7.06e-03, grad_scale: 8.0 2022-12-08 00:59:11,497 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-08 00:59:14,660 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:59:26,099 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:59:39,575 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.154e+02 2.641e+02 3.288e+02 6.152e+02, threshold=5.282e+02, percent-clipped=1.0 2022-12-08 00:59:57,991 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8339, 2.6667, 2.4801, 2.5330, 2.7572, 2.7311, 2.8107, 2.7920], device='cuda:1'), covar=tensor([0.1113, 0.1026, 0.2772, 0.3103, 0.1184, 0.1336, 0.1547, 0.0999], device='cuda:1'), in_proj_covar=tensor([0.0370, 0.0260, 0.0432, 0.0547, 0.0327, 0.0422, 0.0388, 0.0365], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:00:08,003 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:17,815 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82702.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:25,966 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82711.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:26,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0014, 1.8646, 3.2071, 2.3151, 2.9715, 1.9024, 2.4833, 2.9780], device='cuda:1'), covar=tensor([0.0949, 0.4006, 0.0599, 0.4964, 0.0917, 0.2966, 0.1365, 0.0710], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0216, 0.0201, 0.0290, 0.0218, 0.0218, 0.0217, 0.0207], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:00:34,803 INFO [train.py:873] (1/4) Epoch 11, batch 7100, loss[loss=0.1163, simple_loss=0.1541, pruned_loss=0.03925, over 14080.00 frames. ], tot_loss[loss=0.1293, simple_loss=0.1576, pruned_loss=0.0505, over 1876941.10 frames. ], batch size: 26, lr: 7.06e-03, grad_scale: 8.0 2022-12-08 01:01:07,645 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.263e+02 2.729e+02 3.602e+02 7.969e+02, threshold=5.458e+02, percent-clipped=3.0 2022-12-08 01:01:07,747 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82759.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:01:17,422 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.21 vs. limit=2.0 2022-12-08 01:02:00,925 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5963, 1.6011, 1.5674, 1.3153, 1.3059, 1.0215, 1.0958, 1.0712], device='cuda:1'), covar=tensor([0.0172, 0.0278, 0.0181, 0.0207, 0.0275, 0.0414, 0.0244, 0.0451], device='cuda:1'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0015, 0.0015, 0.0025, 0.0020, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:02:03,355 INFO [train.py:873] (1/4) Epoch 11, batch 7200, loss[loss=0.1437, simple_loss=0.1589, pruned_loss=0.06426, over 5940.00 frames. ], tot_loss[loss=0.1303, simple_loss=0.1581, pruned_loss=0.05123, over 1908475.75 frames. ], batch size: 100, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:02:16,364 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2022-12-08 01:02:18,748 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7950, 4.2843, 3.4235, 5.0568, 4.4016, 4.8323, 4.5465, 3.8810], device='cuda:1'), covar=tensor([0.0636, 0.1256, 0.3478, 0.0675, 0.1018, 0.1466, 0.0812, 0.2320], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0296, 0.0269, 0.0255, 0.0309, 0.0294, 0.0251, 0.0252], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-08 01:02:28,444 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82849.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:02:36,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.101e+02 2.276e+02 3.049e+02 3.669e+02 5.112e+02, threshold=6.097e+02, percent-clipped=0.0 2022-12-08 01:03:31,180 INFO [train.py:873] (1/4) Epoch 11, batch 7300, loss[loss=0.1221, simple_loss=0.1498, pruned_loss=0.04719, over 6908.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1575, pruned_loss=0.05018, over 1940792.63 frames. ], batch size: 100, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:03:34,865 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:03:52,523 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=11.12 vs. limit=5.0 2022-12-08 01:04:04,502 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 2.470e+02 3.084e+02 3.689e+02 1.024e+03, threshold=6.168e+02, percent-clipped=3.0 2022-12-08 01:04:08,510 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82963.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:04:36,369 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6489, 1.7603, 1.8192, 1.7172, 1.6809, 0.9949, 1.6076, 1.7354], device='cuda:1'), covar=tensor([0.1049, 0.0947, 0.0521, 0.1107, 0.1359, 0.1109, 0.1137, 0.0877], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0028, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:04:43,211 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:04:56,856 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0966, 2.0794, 2.4073, 1.4817, 1.6327, 2.2258, 1.3223, 2.2291], device='cuda:1'), covar=tensor([0.1202, 0.1659, 0.0945, 0.2350, 0.2923, 0.1051, 0.4129, 0.1066], device='cuda:1'), in_proj_covar=tensor([0.0079, 0.0095, 0.0089, 0.0095, 0.0112, 0.0083, 0.0122, 0.0088], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 01:04:59,089 INFO [train.py:873] (1/4) Epoch 11, batch 7400, loss[loss=0.1168, simple_loss=0.154, pruned_loss=0.03979, over 14224.00 frames. ], tot_loss[loss=0.1284, simple_loss=0.1576, pruned_loss=0.04958, over 1988812.78 frames. ], batch size: 35, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:05:02,187 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83024.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:05:08,448 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-08 01:05:25,167 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83050.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:05:28,388 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-08 01:05:32,577 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 2.140e+02 2.552e+02 3.210e+02 1.054e+03, threshold=5.104e+02, percent-clipped=1.0 2022-12-08 01:05:55,034 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 01:06:19,908 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.43 vs. limit=2.0 2022-12-08 01:06:23,021 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3855, 1.3350, 3.4371, 1.6571, 3.3027, 3.5163, 2.4851, 3.7491], device='cuda:1'), covar=tensor([0.0264, 0.3198, 0.0378, 0.2239, 0.0796, 0.0389, 0.1023, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0158, 0.0160, 0.0170, 0.0171, 0.0174, 0.0134, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 01:06:26,609 INFO [train.py:873] (1/4) Epoch 11, batch 7500, loss[loss=0.09416, simple_loss=0.1391, pruned_loss=0.0246, over 14305.00 frames. ], tot_loss[loss=0.1273, simple_loss=0.1571, pruned_loss=0.0488, over 1994972.77 frames. ], batch size: 25, lr: 7.04e-03, grad_scale: 16.0 2022-12-08 01:06:31,617 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.76 vs. limit=5.0 2022-12-08 01:06:50,736 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83149.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:06:55,368 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 01:06:58,813 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.439e+02 2.915e+02 3.570e+02 7.862e+02, threshold=5.829e+02, percent-clipped=6.0 2022-12-08 01:07:05,295 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4398, 1.0780, 1.2938, 0.8551, 1.1065, 1.4165, 1.1877, 1.1759], device='cuda:1'), covar=tensor([0.0363, 0.0931, 0.0611, 0.0499, 0.0876, 0.0616, 0.0431, 0.1088], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0182, 0.0137, 0.0127, 0.0134, 0.0141, 0.0120, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 01:07:06,744 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0007, 1.9621, 1.6038, 1.9296, 1.8638, 2.0021, 1.8470, 1.7788], device='cuda:1'), covar=tensor([0.0765, 0.0850, 0.2078, 0.0666, 0.0895, 0.0616, 0.1372, 0.1047], device='cuda:1'), in_proj_covar=tensor([0.0266, 0.0296, 0.0268, 0.0254, 0.0309, 0.0295, 0.0253, 0.0251], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-08 01:07:53,460 INFO [train.py:873] (1/4) Epoch 12, batch 0, loss[loss=0.1149, simple_loss=0.156, pruned_loss=0.03691, over 14209.00 frames. ], tot_loss[loss=0.1149, simple_loss=0.156, pruned_loss=0.03691, over 14209.00 frames. ], batch size: 35, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:07:53,461 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 01:08:00,688 INFO [train.py:905] (1/4) Epoch 12, validation: loss=0.1326, simple_loss=0.1738, pruned_loss=0.04568, over 857387.00 frames. 2022-12-08 01:08:00,688 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 01:08:12,340 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83196.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:08:13,520 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83197.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:08:38,648 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:08:54,783 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 01:09:07,377 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83257.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:08,850 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.351e+01 1.849e+02 2.933e+02 3.942e+02 9.490e+02, threshold=5.866e+02, percent-clipped=6.0 2022-12-08 01:09:21,074 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:30,010 INFO [train.py:873] (1/4) Epoch 12, batch 100, loss[loss=0.1509, simple_loss=0.1694, pruned_loss=0.06623, over 13528.00 frames. ], tot_loss[loss=0.1309, simple_loss=0.1594, pruned_loss=0.05123, over 870047.99 frames. ], batch size: 100, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:09:30,979 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:34,427 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3438, 3.0635, 2.9790, 2.1034, 2.7976, 3.1070, 3.3615, 2.6666], device='cuda:1'), covar=tensor([0.0687, 0.1162, 0.1027, 0.1672, 0.1073, 0.0744, 0.0646, 0.1242], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0185, 0.0139, 0.0128, 0.0136, 0.0143, 0.0121, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 01:09:53,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9439, 1.9697, 4.5685, 4.2343, 4.1649, 4.7300, 4.3163, 4.7469], device='cuda:1'), covar=tensor([0.1551, 0.1373, 0.0099, 0.0176, 0.0194, 0.0094, 0.0122, 0.0101], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0129, 0.0167, 0.0144, 0.0138, 0.0120, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:10:00,906 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:10:24,255 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83345.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:10:36,278 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.243e+02 2.760e+02 3.223e+02 5.001e+02, threshold=5.521e+02, percent-clipped=0.0 2022-12-08 01:10:57,432 INFO [train.py:873] (1/4) Epoch 12, batch 200, loss[loss=0.1345, simple_loss=0.1618, pruned_loss=0.05355, over 13530.00 frames. ], tot_loss[loss=0.1295, simple_loss=0.1587, pruned_loss=0.05019, over 1372961.78 frames. ], batch size: 100, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:11:45,901 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2346, 3.9591, 3.8589, 4.2903, 3.9806, 3.7651, 4.2673, 3.6392], device='cuda:1'), covar=tensor([0.0493, 0.1029, 0.0416, 0.0466, 0.0889, 0.0965, 0.0645, 0.0565], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0254, 0.0180, 0.0177, 0.0172, 0.0144, 0.0266, 0.0158], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 01:12:04,537 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.214e+02 2.674e+02 3.340e+02 6.097e+02, threshold=5.348e+02, percent-clipped=3.0 2022-12-08 01:12:20,088 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6759, 2.0304, 2.5419, 2.6793, 2.6477, 2.0476, 2.6318, 2.2379], device='cuda:1'), covar=tensor([0.0383, 0.0750, 0.0605, 0.0417, 0.0393, 0.1044, 0.0366, 0.0657], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0251, 0.0371, 0.0319, 0.0261, 0.0298, 0.0299, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:12:24,713 INFO [train.py:873] (1/4) Epoch 12, batch 300, loss[loss=0.1375, simple_loss=0.1708, pruned_loss=0.05204, over 14553.00 frames. ], tot_loss[loss=0.1285, simple_loss=0.1577, pruned_loss=0.04963, over 1623773.72 frames. ], batch size: 43, lr: 6.73e-03, grad_scale: 8.0 2022-12-08 01:13:25,082 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9894, 3.6630, 2.8673, 4.3980, 4.0792, 4.0998, 3.5998, 2.9669], device='cuda:1'), covar=tensor([0.0954, 0.1395, 0.3786, 0.0552, 0.0940, 0.1556, 0.1231, 0.3517], device='cuda:1'), in_proj_covar=tensor([0.0267, 0.0294, 0.0267, 0.0255, 0.0308, 0.0295, 0.0252, 0.0250], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-08 01:13:25,766 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83552.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:13:32,202 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.254e+02 2.674e+02 3.399e+02 5.874e+02, threshold=5.349e+02, percent-clipped=4.0 2022-12-08 01:13:43,019 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2022-12-08 01:13:45,352 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 01:13:49,056 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 01:13:53,020 INFO [train.py:873] (1/4) Epoch 12, batch 400, loss[loss=0.1409, simple_loss=0.1512, pruned_loss=0.0653, over 3876.00 frames. ], tot_loss[loss=0.1271, simple_loss=0.157, pruned_loss=0.04864, over 1776776.93 frames. ], batch size: 100, lr: 6.73e-03, grad_scale: 8.0 2022-12-08 01:14:12,392 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.84 vs. limit=5.0 2022-12-08 01:14:25,420 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83619.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:14:27,345 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8590, 3.6107, 3.5893, 3.9215, 3.5723, 3.3574, 3.9148, 3.7647], device='cuda:1'), covar=tensor([0.0660, 0.0918, 0.0797, 0.0608, 0.0784, 0.0719, 0.0565, 0.0757], device='cuda:1'), in_proj_covar=tensor([0.0131, 0.0128, 0.0135, 0.0143, 0.0136, 0.0112, 0.0154, 0.0133], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 01:14:43,829 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83640.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:14:47,467 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.12 vs. limit=2.0 2022-12-08 01:15:01,084 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.765e+01 2.109e+02 2.628e+02 3.451e+02 6.835e+02, threshold=5.257e+02, percent-clipped=4.0 2022-12-08 01:15:07,459 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83667.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:15:21,383 INFO [train.py:873] (1/4) Epoch 12, batch 500, loss[loss=0.1534, simple_loss=0.1682, pruned_loss=0.06925, over 6928.00 frames. ], tot_loss[loss=0.1275, simple_loss=0.1571, pruned_loss=0.049, over 1876376.38 frames. ], batch size: 100, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:15:47,386 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0160, 2.4669, 3.7861, 2.8035, 3.8082, 3.6860, 3.5811, 3.1961], device='cuda:1'), covar=tensor([0.0645, 0.2975, 0.0846, 0.1845, 0.0793, 0.0842, 0.1378, 0.1646], device='cuda:1'), in_proj_covar=tensor([0.0347, 0.0312, 0.0392, 0.0300, 0.0371, 0.0318, 0.0363, 0.0309], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:16:28,955 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.083e+02 2.759e+02 3.343e+02 7.485e+02, threshold=5.518e+02, percent-clipped=2.0 2022-12-08 01:16:29,205 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83760.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:16:49,377 INFO [train.py:873] (1/4) Epoch 12, batch 600, loss[loss=0.1172, simple_loss=0.1152, pruned_loss=0.05958, over 1240.00 frames. ], tot_loss[loss=0.1273, simple_loss=0.1569, pruned_loss=0.04887, over 1859902.88 frames. ], batch size: 100, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:17:22,702 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:17:50,201 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:17:53,042 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7882, 2.9558, 4.6534, 3.4331, 4.5514, 4.3298, 4.2641, 3.8973], device='cuda:1'), covar=tensor([0.0736, 0.3179, 0.0961, 0.2001, 0.0835, 0.0860, 0.2002, 0.2070], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0314, 0.0390, 0.0299, 0.0373, 0.0318, 0.0363, 0.0309], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:17:53,325 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-08 01:17:57,121 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.241e+02 2.685e+02 3.305e+02 1.014e+03, threshold=5.370e+02, percent-clipped=1.0 2022-12-08 01:18:17,198 INFO [train.py:873] (1/4) Epoch 12, batch 700, loss[loss=0.1205, simple_loss=0.1544, pruned_loss=0.04332, over 14266.00 frames. ], tot_loss[loss=0.1266, simple_loss=0.1563, pruned_loss=0.0485, over 1904940.43 frames. ], batch size: 80, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:18:28,338 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2022-12-08 01:18:32,214 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83900.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:06,824 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83940.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:14,705 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83949.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:23,841 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.241e+02 2.717e+02 3.335e+02 9.146e+02, threshold=5.433e+02, percent-clipped=9.0 2022-12-08 01:19:44,262 INFO [train.py:873] (1/4) Epoch 12, batch 800, loss[loss=0.1002, simple_loss=0.1324, pruned_loss=0.03396, over 14353.00 frames. ], tot_loss[loss=0.1283, simple_loss=0.1569, pruned_loss=0.04987, over 1854244.49 frames. ], batch size: 18, lr: 6.71e-03, grad_scale: 8.0 2022-12-08 01:19:46,292 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:48,763 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83988.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:58,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7408, 0.8013, 0.8093, 0.8772, 0.8336, 0.3332, 0.7411, 0.8900], device='cuda:1'), covar=tensor([0.0237, 0.0302, 0.0413, 0.0257, 0.0307, 0.0251, 0.0581, 0.0425], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:20:08,153 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84010.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 01:20:13,580 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7355, 1.3768, 3.7200, 1.7313, 3.7299, 3.8857, 2.7802, 4.1438], device='cuda:1'), covar=tensor([0.0252, 0.3305, 0.0458, 0.2335, 0.0578, 0.0449, 0.0985, 0.0196], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0158, 0.0161, 0.0171, 0.0170, 0.0174, 0.0135, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:20:26,658 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6054, 1.4121, 3.6196, 1.6029, 3.5650, 3.7110, 2.6643, 3.9617], device='cuda:1'), covar=tensor([0.0232, 0.3062, 0.0404, 0.2346, 0.0599, 0.0379, 0.0738, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0158, 0.0161, 0.0170, 0.0170, 0.0174, 0.0135, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:20:29,279 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5357, 2.3767, 3.0105, 2.0002, 1.9620, 2.6042, 1.5243, 2.5137], device='cuda:1'), covar=tensor([0.1099, 0.1610, 0.0764, 0.2072, 0.2724, 0.1020, 0.4094, 0.1031], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0094, 0.0088, 0.0094, 0.0112, 0.0081, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 01:20:40,366 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:20:52,176 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.165e+02 2.773e+02 3.425e+02 7.521e+02, threshold=5.546e+02, percent-clipped=4.0 2022-12-08 01:21:11,643 INFO [train.py:873] (1/4) Epoch 12, batch 900, loss[loss=0.1307, simple_loss=0.1564, pruned_loss=0.05253, over 10328.00 frames. ], tot_loss[loss=0.1284, simple_loss=0.1571, pruned_loss=0.04984, over 1924572.54 frames. ], batch size: 100, lr: 6.71e-03, grad_scale: 8.0 2022-12-08 01:21:24,355 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:21:37,095 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5816, 2.2753, 2.4926, 1.6768, 2.1604, 2.4384, 2.5752, 2.1751], device='cuda:1'), covar=tensor([0.0751, 0.0901, 0.0908, 0.1479, 0.1163, 0.0743, 0.0576, 0.1367], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0180, 0.0136, 0.0125, 0.0134, 0.0142, 0.0120, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 01:21:40,415 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:22:17,036 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:22:18,574 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.043e+02 2.201e+02 2.633e+02 3.303e+02 7.770e+02, threshold=5.265e+02, percent-clipped=5.0 2022-12-08 01:22:32,248 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9945, 2.0424, 2.8568, 2.2879, 2.8124, 2.8304, 2.6418, 2.3660], device='cuda:1'), covar=tensor([0.0692, 0.2591, 0.0760, 0.1591, 0.0613, 0.0804, 0.0828, 0.1636], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0315, 0.0394, 0.0300, 0.0373, 0.0317, 0.0367, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:22:36,299 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1462, 3.8944, 3.8239, 4.1880, 3.9207, 3.6676, 4.2008, 3.5581], device='cuda:1'), covar=tensor([0.0449, 0.0823, 0.0401, 0.0437, 0.0788, 0.1369, 0.0525, 0.0557], device='cuda:1'), in_proj_covar=tensor([0.0169, 0.0258, 0.0182, 0.0182, 0.0175, 0.0146, 0.0270, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 01:22:38,829 INFO [train.py:873] (1/4) Epoch 12, batch 1000, loss[loss=0.1447, simple_loss=0.1657, pruned_loss=0.06186, over 14330.00 frames. ], tot_loss[loss=0.1281, simple_loss=0.1572, pruned_loss=0.04949, over 1960660.39 frames. ], batch size: 55, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:22:52,889 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84199.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:22:57,989 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8216, 3.5934, 3.6070, 3.8578, 3.3775, 3.2650, 3.8586, 3.7368], device='cuda:1'), covar=tensor([0.0630, 0.0864, 0.0715, 0.0603, 0.0861, 0.0752, 0.0650, 0.0710], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0128, 0.0136, 0.0145, 0.0137, 0.0113, 0.0156, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 01:23:25,620 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84236.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:23:25,629 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8346, 0.8444, 0.7604, 0.8701, 0.8433, 0.3082, 0.6431, 0.9164], device='cuda:1'), covar=tensor([0.0239, 0.0497, 0.0452, 0.0292, 0.0317, 0.0285, 0.0613, 0.0537], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0028, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:23:46,259 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.176e+02 2.741e+02 3.221e+02 6.313e+02, threshold=5.481e+02, percent-clipped=1.0 2022-12-08 01:23:46,476 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:06,155 INFO [train.py:873] (1/4) Epoch 12, batch 1100, loss[loss=0.1252, simple_loss=0.1583, pruned_loss=0.04608, over 11160.00 frames. ], tot_loss[loss=0.1266, simple_loss=0.1564, pruned_loss=0.04842, over 1968118.09 frames. ], batch size: 100, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:24:11,971 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0069, 2.9399, 2.2086, 3.1174, 2.9275, 2.9450, 2.6529, 2.2841], device='cuda:1'), covar=tensor([0.1025, 0.1441, 0.3167, 0.0733, 0.0908, 0.0937, 0.1316, 0.3192], device='cuda:1'), in_proj_covar=tensor([0.0264, 0.0289, 0.0263, 0.0253, 0.0304, 0.0292, 0.0250, 0.0247], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:24:18,772 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84297.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:25,567 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84305.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:24:52,910 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:56,767 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84341.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:25:13,032 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.219e+02 2.734e+02 3.210e+02 6.975e+02, threshold=5.467e+02, percent-clipped=2.0 2022-12-08 01:25:24,464 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9905, 1.1864, 1.0449, 1.0668, 1.0176, 0.6270, 0.9302, 1.1662], device='cuda:1'), covar=tensor([0.0360, 0.0439, 0.0391, 0.0413, 0.0415, 0.0388, 0.0729, 0.0597], device='cuda:1'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:25:27,629 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 01:25:33,227 INFO [train.py:873] (1/4) Epoch 12, batch 1200, loss[loss=0.1677, simple_loss=0.1559, pruned_loss=0.08978, over 1305.00 frames. ], tot_loss[loss=0.126, simple_loss=0.1562, pruned_loss=0.04785, over 1969527.58 frames. ], batch size: 100, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:25:45,502 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84397.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:02,530 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84416.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:21,998 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9055, 1.5586, 3.5325, 3.2238, 3.3334, 3.5078, 2.9836, 3.5725], device='cuda:1'), covar=tensor([0.1423, 0.1502, 0.0127, 0.0272, 0.0254, 0.0147, 0.0268, 0.0126], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0129, 0.0165, 0.0144, 0.0138, 0.0120, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:26:23,994 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2022-12-08 01:26:34,076 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84453.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:40,233 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.298e+02 2.755e+02 3.792e+02 6.830e+02, threshold=5.510e+02, percent-clipped=5.0 2022-12-08 01:26:43,675 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84464.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:50,699 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2022-12-08 01:27:00,263 INFO [train.py:873] (1/4) Epoch 12, batch 1300, loss[loss=0.1717, simple_loss=0.179, pruned_loss=0.08225, over 8584.00 frames. ], tot_loss[loss=0.1264, simple_loss=0.1563, pruned_loss=0.04821, over 1980011.12 frames. ], batch size: 100, lr: 6.69e-03, grad_scale: 8.0 2022-12-08 01:27:36,947 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84525.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:03,423 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:07,761 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 2.228e+02 2.625e+02 3.303e+02 8.483e+02, threshold=5.249e+02, percent-clipped=5.0 2022-12-08 01:28:27,680 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6165, 4.6236, 4.9916, 3.9433, 4.7559, 5.0427, 1.8926, 4.4303], device='cuda:1'), covar=tensor([0.0205, 0.0324, 0.0283, 0.0480, 0.0282, 0.0181, 0.2917, 0.0252], device='cuda:1'), in_proj_covar=tensor([0.0158, 0.0167, 0.0140, 0.0136, 0.0200, 0.0135, 0.0156, 0.0184], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:28:28,445 INFO [train.py:873] (1/4) Epoch 12, batch 1400, loss[loss=0.09915, simple_loss=0.144, pruned_loss=0.02714, over 14260.00 frames. ], tot_loss[loss=0.1262, simple_loss=0.1561, pruned_loss=0.04812, over 1938053.08 frames. ], batch size: 28, lr: 6.69e-03, grad_scale: 8.0 2022-12-08 01:28:31,079 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84586.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:36,443 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84592.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:48,177 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:12,469 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84633.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:14,680 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 01:29:19,462 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:26,528 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9898, 2.0430, 3.0322, 3.0716, 2.9593, 2.2187, 2.9771, 2.4039], device='cuda:1'), covar=tensor([0.0370, 0.0819, 0.0604, 0.0405, 0.0388, 0.1173, 0.0360, 0.0774], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0250, 0.0370, 0.0319, 0.0259, 0.0296, 0.0297, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 01:29:29,690 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:35,866 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.687e+01 2.089e+02 2.664e+02 3.503e+02 6.178e+02, threshold=5.327e+02, percent-clipped=6.0 2022-12-08 01:29:55,407 INFO [train.py:873] (1/4) Epoch 12, batch 1500, loss[loss=0.1275, simple_loss=0.14, pruned_loss=0.05748, over 3852.00 frames. ], tot_loss[loss=0.1258, simple_loss=0.1555, pruned_loss=0.04807, over 1901510.41 frames. ], batch size: 100, lr: 6.68e-03, grad_scale: 8.0 2022-12-08 01:30:00,987 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84689.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:02,448 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.02 vs. limit=5.0 2022-12-08 01:30:03,575 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:04,761 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2022-12-08 01:30:05,340 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84694.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:07,054 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 01:30:56,781 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84753.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:00,898 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.82 vs. limit=5.0 2022-12-08 01:31:02,108 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.424e+02 2.855e+02 3.432e+02 8.447e+02, threshold=5.711e+02, percent-clipped=4.0 2022-12-08 01:31:03,107 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0817, 2.9009, 2.5679, 2.7492, 2.9751, 2.9955, 3.0375, 3.0400], device='cuda:1'), covar=tensor([0.0946, 0.0633, 0.2261, 0.2419, 0.0818, 0.0955, 0.1091, 0.0842], device='cuda:1'), in_proj_covar=tensor([0.0368, 0.0254, 0.0429, 0.0547, 0.0321, 0.0418, 0.0388, 0.0363], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:31:22,262 INFO [train.py:873] (1/4) Epoch 12, batch 1600, loss[loss=0.1442, simple_loss=0.1625, pruned_loss=0.06295, over 7804.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1549, pruned_loss=0.04766, over 1988827.63 frames. ], batch size: 100, lr: 6.68e-03, grad_scale: 8.0 2022-12-08 01:31:36,890 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84800.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:37,627 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84801.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:38,961 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-08 01:32:25,413 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84855.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:29,445 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.720e+01 2.080e+02 2.547e+02 3.159e+02 6.486e+02, threshold=5.095e+02, percent-clipped=3.0 2022-12-08 01:32:30,498 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84861.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:37,535 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.46 vs. limit=2.0 2022-12-08 01:32:47,203 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:48,864 INFO [train.py:873] (1/4) Epoch 12, batch 1700, loss[loss=0.0977, simple_loss=0.1309, pruned_loss=0.03227, over 13664.00 frames. ], tot_loss[loss=0.1278, simple_loss=0.1568, pruned_loss=0.04938, over 1942207.26 frames. ], batch size: 17, lr: 6.68e-03, grad_scale: 4.0 2022-12-08 01:32:57,337 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:06,797 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84903.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:39,686 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84940.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:49,335 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84951.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:57,762 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.143e+02 2.620e+02 3.286e+02 6.897e+02, threshold=5.240e+02, percent-clipped=6.0 2022-12-08 01:34:17,444 INFO [train.py:873] (1/4) Epoch 12, batch 1800, loss[loss=0.1956, simple_loss=0.1746, pruned_loss=0.1082, over 1212.00 frames. ], tot_loss[loss=0.1279, simple_loss=0.1571, pruned_loss=0.04935, over 1926243.91 frames. ], batch size: 100, lr: 6.67e-03, grad_scale: 4.0 2022-12-08 01:34:22,794 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:34:25,414 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:34:47,146 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85012.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:35:10,622 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:35:29,534 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.196e+02 2.693e+02 3.379e+02 9.036e+02, threshold=5.386e+02, percent-clipped=1.0 2022-12-08 01:35:48,290 INFO [train.py:873] (1/4) Epoch 12, batch 1900, loss[loss=0.1349, simple_loss=0.1639, pruned_loss=0.05293, over 14282.00 frames. ], tot_loss[loss=0.1293, simple_loss=0.1583, pruned_loss=0.05021, over 1942168.97 frames. ], batch size: 35, lr: 6.67e-03, grad_scale: 4.0 2022-12-08 01:36:09,487 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85107.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:16,213 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:18,343 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7325, 1.5658, 4.0464, 3.7684, 3.8489, 4.1998, 3.6382, 4.1127], device='cuda:1'), covar=tensor([0.1838, 0.1688, 0.0187, 0.0292, 0.0263, 0.0195, 0.0234, 0.0209], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0155, 0.0127, 0.0164, 0.0142, 0.0136, 0.0118, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:36:31,706 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6220, 1.5192, 1.6582, 1.4912, 1.3821, 1.2237, 1.2112, 1.0679], device='cuda:1'), covar=tensor([0.0153, 0.0209, 0.0190, 0.0159, 0.0247, 0.0310, 0.0262, 0.0384], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0017, 0.0015, 0.0016, 0.0015, 0.0026, 0.0021, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:36:52,429 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85156.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:56,465 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.007e+02 2.116e+02 2.633e+02 3.281e+02 1.062e+03, threshold=5.267e+02, percent-clipped=7.0 2022-12-08 01:36:56,658 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1962, 1.8328, 2.2644, 1.5049, 1.8489, 2.1965, 2.1536, 1.9476], device='cuda:1'), covar=tensor([0.0935, 0.0939, 0.1041, 0.1638, 0.1694, 0.0843, 0.0816, 0.1801], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0184, 0.0139, 0.0127, 0.0137, 0.0145, 0.0123, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 01:37:00,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3276, 2.8090, 4.1125, 3.1655, 4.1090, 4.0880, 3.9113, 3.4233], device='cuda:1'), covar=tensor([0.0776, 0.2743, 0.0963, 0.1584, 0.0827, 0.0754, 0.1731, 0.1772], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0312, 0.0397, 0.0300, 0.0370, 0.0318, 0.0365, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:37:03,254 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85168.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:37:10,136 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:37:14,095 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85181.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:37:15,662 INFO [train.py:873] (1/4) Epoch 12, batch 2000, loss[loss=0.1339, simple_loss=0.1634, pruned_loss=0.05215, over 11174.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1581, pruned_loss=0.04983, over 1975503.45 frames. ], batch size: 100, lr: 6.66e-03, grad_scale: 8.0 2022-12-08 01:37:40,011 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9192, 2.2307, 3.8876, 4.0464, 3.8915, 2.2549, 4.0435, 3.1764], device='cuda:1'), covar=tensor([0.0366, 0.0930, 0.0733, 0.0381, 0.0346, 0.1505, 0.0349, 0.0778], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0253, 0.0370, 0.0320, 0.0261, 0.0299, 0.0300, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:37:56,096 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:38:25,282 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.573e+01 2.203e+02 2.703e+02 3.626e+02 1.921e+03, threshold=5.405e+02, percent-clipped=8.0 2022-12-08 01:38:32,304 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.0040, 5.4552, 5.3264, 5.9244, 5.4827, 4.8310, 5.9373, 4.7394], device='cuda:1'), covar=tensor([0.0298, 0.0894, 0.0305, 0.0436, 0.0774, 0.0343, 0.0471, 0.0548], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0264, 0.0185, 0.0182, 0.0177, 0.0150, 0.0273, 0.0162], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 01:38:43,506 INFO [train.py:873] (1/4) Epoch 12, batch 2100, loss[loss=0.15, simple_loss=0.1448, pruned_loss=0.0776, over 1291.00 frames. ], tot_loss[loss=0.1273, simple_loss=0.157, pruned_loss=0.04884, over 1974514.52 frames. ], batch size: 100, lr: 6.66e-03, grad_scale: 4.0 2022-12-08 01:38:48,961 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:39:04,429 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:39:08,117 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 01:39:30,446 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:39:42,927 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2022-12-08 01:39:51,892 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.818e+01 2.339e+02 2.945e+02 3.643e+02 8.230e+02, threshold=5.891e+02, percent-clipped=7.0 2022-12-08 01:40:10,081 INFO [train.py:873] (1/4) Epoch 12, batch 2200, loss[loss=0.1257, simple_loss=0.1534, pruned_loss=0.049, over 11133.00 frames. ], tot_loss[loss=0.1288, simple_loss=0.1579, pruned_loss=0.04991, over 1929132.22 frames. ], batch size: 100, lr: 6.66e-03, grad_scale: 4.0 2022-12-08 01:40:35,791 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 01:40:49,609 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:14,534 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85456.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:19,365 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.150e+02 2.815e+02 3.417e+02 5.996e+02, threshold=5.631e+02, percent-clipped=2.0 2022-12-08 01:41:20,083 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85463.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 01:41:26,834 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:31,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9198, 2.9630, 3.0991, 2.9556, 2.9869, 2.8033, 1.4588, 2.7439], device='cuda:1'), covar=tensor([0.0397, 0.0465, 0.0449, 0.0463, 0.0425, 0.0875, 0.2825, 0.0369], device='cuda:1'), in_proj_covar=tensor([0.0161, 0.0171, 0.0141, 0.0139, 0.0203, 0.0136, 0.0158, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 01:41:36,989 INFO [train.py:873] (1/4) Epoch 12, batch 2300, loss[loss=0.182, simple_loss=0.1621, pruned_loss=0.101, over 1297.00 frames. ], tot_loss[loss=0.1279, simple_loss=0.1571, pruned_loss=0.04939, over 1933799.97 frames. ], batch size: 100, lr: 6.65e-03, grad_scale: 4.0 2022-12-08 01:41:37,186 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:37,943 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8411, 1.4722, 1.8131, 1.2760, 1.4921, 1.8637, 1.6651, 1.5417], device='cuda:1'), covar=tensor([0.0678, 0.0750, 0.0810, 0.0987, 0.1342, 0.1005, 0.0602, 0.1545], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0183, 0.0138, 0.0126, 0.0136, 0.0144, 0.0121, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 01:41:42,381 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:55,417 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85504.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:18,060 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85530.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:30,179 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85544.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:34,837 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 01:42:40,890 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6366, 1.7707, 1.9104, 1.2691, 1.2866, 1.7179, 1.1219, 1.7769], device='cuda:1'), covar=tensor([0.1146, 0.1946, 0.0666, 0.1926, 0.2718, 0.0860, 0.3252, 0.0929], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0098, 0.0091, 0.0096, 0.0116, 0.0084, 0.0124, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 01:42:45,171 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.375e+02 2.779e+02 3.534e+02 5.701e+02, threshold=5.557e+02, percent-clipped=1.0 2022-12-08 01:43:03,727 INFO [train.py:873] (1/4) Epoch 12, batch 2400, loss[loss=0.1144, simple_loss=0.1528, pruned_loss=0.03795, over 14064.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1577, pruned_loss=0.05006, over 1956811.90 frames. ], batch size: 29, lr: 6.65e-03, grad_scale: 8.0 2022-12-08 01:43:09,643 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8861, 1.7770, 3.1041, 2.2776, 2.9537, 1.9101, 2.4694, 2.9271], device='cuda:1'), covar=tensor([0.1045, 0.4267, 0.0610, 0.4638, 0.0847, 0.3356, 0.1214, 0.0627], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0215, 0.0203, 0.0287, 0.0223, 0.0218, 0.0215, 0.0208], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:43:10,438 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:43:14,311 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2022-12-08 01:43:24,430 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85607.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:43:34,335 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0761, 1.9879, 2.2167, 1.3251, 1.5856, 2.0765, 1.2243, 2.0449], device='cuda:1'), covar=tensor([0.0882, 0.1649, 0.0825, 0.2819, 0.3242, 0.0899, 0.3872, 0.1083], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0098, 0.0090, 0.0096, 0.0115, 0.0084, 0.0124, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 01:43:40,318 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:44:06,764 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85655.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:44:08,567 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8853, 1.2340, 2.0627, 1.3415, 1.9814, 2.1087, 1.7220, 2.1410], device='cuda:1'), covar=tensor([0.0300, 0.1949, 0.0392, 0.1574, 0.0544, 0.0473, 0.0994, 0.0325], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0158, 0.0160, 0.0168, 0.0168, 0.0173, 0.0134, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 01:44:11,136 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8704, 1.4721, 2.8401, 2.5629, 2.7366, 2.8337, 2.1234, 2.8051], device='cuda:1'), covar=tensor([0.0920, 0.1141, 0.0129, 0.0342, 0.0307, 0.0158, 0.0466, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0128, 0.0168, 0.0144, 0.0139, 0.0120, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:44:12,664 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.042e+02 2.557e+02 3.044e+02 5.063e+02, threshold=5.114e+02, percent-clipped=0.0 2022-12-08 01:44:30,324 INFO [train.py:873] (1/4) Epoch 12, batch 2500, loss[loss=0.1749, simple_loss=0.1838, pruned_loss=0.08304, over 7814.00 frames. ], tot_loss[loss=0.1286, simple_loss=0.1574, pruned_loss=0.04991, over 1926232.36 frames. ], batch size: 100, lr: 6.65e-03, grad_scale: 8.0 2022-12-08 01:44:33,076 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85686.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:44:48,281 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-08 01:45:38,819 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 01:45:40,094 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.190e+02 2.751e+02 3.272e+02 6.413e+02, threshold=5.502e+02, percent-clipped=4.0 2022-12-08 01:45:40,302 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:45:47,389 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:45:58,010 INFO [train.py:873] (1/4) Epoch 12, batch 2600, loss[loss=0.1252, simple_loss=0.1518, pruned_loss=0.04929, over 6030.00 frames. ], tot_loss[loss=0.1274, simple_loss=0.157, pruned_loss=0.0489, over 1991799.59 frames. ], batch size: 100, lr: 6.64e-03, grad_scale: 4.0 2022-12-08 01:45:58,978 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:10,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 01:46:22,756 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85811.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:29,916 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:46,842 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85839.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:47:08,283 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.414e+01 2.205e+02 2.825e+02 3.419e+02 7.121e+02, threshold=5.651e+02, percent-clipped=1.0 2022-12-08 01:47:18,430 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:47:25,017 INFO [train.py:873] (1/4) Epoch 12, batch 2700, loss[loss=0.1189, simple_loss=0.1337, pruned_loss=0.052, over 3849.00 frames. ], tot_loss[loss=0.1265, simple_loss=0.1561, pruned_loss=0.04845, over 1923190.66 frames. ], batch size: 100, lr: 6.64e-03, grad_scale: 4.0 2022-12-08 01:47:27,750 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85886.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:47:45,132 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 01:48:06,406 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85930.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:48:11,737 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:48:29,695 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85957.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:48:34,526 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.895e+01 2.358e+02 3.006e+02 3.476e+02 6.598e+02, threshold=6.011e+02, percent-clipped=3.0 2022-12-08 01:48:51,190 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85981.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:48:52,815 INFO [train.py:873] (1/4) Epoch 12, batch 2800, loss[loss=0.1333, simple_loss=0.163, pruned_loss=0.05181, over 14241.00 frames. ], tot_loss[loss=0.127, simple_loss=0.1566, pruned_loss=0.0487, over 1938990.12 frames. ], batch size: 69, lr: 6.63e-03, grad_scale: 8.0 2022-12-08 01:48:59,857 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85991.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:49:09,080 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.67 vs. limit=5.0 2022-12-08 01:49:09,703 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:49:24,388 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:49:28,768 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4013, 2.2265, 3.1237, 2.4847, 3.1680, 3.1189, 2.9340, 2.6116], device='cuda:1'), covar=tensor([0.0855, 0.2853, 0.0972, 0.1916, 0.0735, 0.0876, 0.1201, 0.1951], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0314, 0.0397, 0.0303, 0.0375, 0.0319, 0.0370, 0.0309], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:49:53,794 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86052.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:03,560 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.107e+02 2.317e+02 2.627e+02 3.139e+02 9.999e+02, threshold=5.255e+02, percent-clipped=3.0 2022-12-08 01:50:03,800 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:08,731 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86069.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:20,497 INFO [train.py:873] (1/4) Epoch 12, batch 2900, loss[loss=0.1119, simple_loss=0.1509, pruned_loss=0.03648, over 14486.00 frames. ], tot_loss[loss=0.1259, simple_loss=0.1556, pruned_loss=0.0481, over 1927081.02 frames. ], batch size: 34, lr: 6.63e-03, grad_scale: 8.0 2022-12-08 01:50:21,276 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:46,753 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86113.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:01,277 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:03,170 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:08,971 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6824, 3.3675, 3.3202, 3.6466, 3.3869, 3.6148, 3.6630, 3.0342], device='cuda:1'), covar=tensor([0.0463, 0.1063, 0.0481, 0.0528, 0.0866, 0.0356, 0.0649, 0.0616], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0266, 0.0189, 0.0183, 0.0180, 0.0151, 0.0276, 0.0162], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 01:51:09,837 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86139.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:30,509 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.360e+01 2.186e+02 2.706e+02 3.386e+02 6.755e+02, threshold=5.412e+02, percent-clipped=5.0 2022-12-08 01:51:47,631 INFO [train.py:873] (1/4) Epoch 12, batch 3000, loss[loss=0.127, simple_loss=0.1617, pruned_loss=0.04617, over 14269.00 frames. ], tot_loss[loss=0.1262, simple_loss=0.1557, pruned_loss=0.04836, over 1909902.68 frames. ], batch size: 60, lr: 6.63e-03, grad_scale: 4.0 2022-12-08 01:51:47,631 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 01:51:53,251 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3249, 4.3526, 4.2193, 4.4427, 3.9888, 3.9383, 4.4267, 4.1674], device='cuda:1'), covar=tensor([0.0538, 0.0334, 0.0700, 0.0574, 0.0846, 0.0332, 0.0378, 0.0760], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0128, 0.0136, 0.0147, 0.0137, 0.0112, 0.0154, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 01:51:56,041 INFO [train.py:905] (1/4) Epoch 12, validation: loss=0.1299, simple_loss=0.1698, pruned_loss=0.04501, over 857387.00 frames. 2022-12-08 01:51:56,041 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 01:51:58,751 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86186.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:59,511 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86187.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:03,746 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-08 01:52:38,280 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86231.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:41,123 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:46,097 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9276, 3.7281, 3.5992, 4.0076, 3.7612, 3.4482, 3.9910, 3.3492], device='cuda:1'), covar=tensor([0.0648, 0.1052, 0.0458, 0.0526, 0.0835, 0.1621, 0.0634, 0.0637], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0266, 0.0188, 0.0183, 0.0179, 0.0150, 0.0275, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 01:52:52,164 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86246.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:07,352 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.699e+01 2.315e+02 2.814e+02 3.910e+02 8.994e+02, threshold=5.629e+02, percent-clipped=7.0 2022-12-08 01:53:11,932 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1233, 1.7595, 4.6409, 4.2450, 4.0865, 4.7132, 4.3247, 4.7809], device='cuda:1'), covar=tensor([0.1417, 0.1478, 0.0085, 0.0176, 0.0215, 0.0106, 0.0150, 0.0093], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0157, 0.0127, 0.0166, 0.0144, 0.0140, 0.0119, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 01:53:23,059 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:24,629 INFO [train.py:873] (1/4) Epoch 12, batch 3100, loss[loss=0.1022, simple_loss=0.1424, pruned_loss=0.03096, over 14660.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1544, pruned_loss=0.04651, over 1953570.48 frames. ], batch size: 22, lr: 6.62e-03, grad_scale: 4.0 2022-12-08 01:53:27,167 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86286.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:53:45,085 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:50,973 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86313.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:04,960 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:31,028 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86358.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:36,315 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.157e+02 2.639e+02 3.347e+02 7.885e+02, threshold=5.278e+02, percent-clipped=4.0 2022-12-08 01:54:43,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6691, 3.0139, 4.2857, 3.2768, 4.2662, 4.2739, 4.0525, 3.8590], device='cuda:1'), covar=tensor([0.0621, 0.2812, 0.1007, 0.1893, 0.0868, 0.0806, 0.1756, 0.1600], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0313, 0.0398, 0.0305, 0.0377, 0.0319, 0.0369, 0.0311], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:54:51,745 INFO [train.py:873] (1/4) Epoch 12, batch 3200, loss[loss=0.1463, simple_loss=0.1618, pruned_loss=0.06542, over 6969.00 frames. ], tot_loss[loss=0.1243, simple_loss=0.1551, pruned_loss=0.04674, over 2067545.83 frames. ], batch size: 100, lr: 6.62e-03, grad_scale: 8.0 2022-12-08 01:55:03,981 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6753, 1.5393, 1.5666, 1.6874, 1.6343, 1.0679, 1.3436, 1.5036], device='cuda:1'), covar=tensor([0.0571, 0.0792, 0.0736, 0.0490, 0.0376, 0.0777, 0.0978, 0.0726], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0027, 0.0030, 0.0026, 0.0029, 0.0039, 0.0028, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:55:14,589 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:28,827 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:30,577 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4210, 1.4649, 1.5937, 1.2283, 1.1778, 1.1244, 1.0194, 1.0339], device='cuda:1'), covar=tensor([0.0180, 0.0268, 0.0177, 0.0244, 0.0233, 0.0389, 0.0239, 0.0531], device='cuda:1'), in_proj_covar=tensor([0.0016, 0.0017, 0.0015, 0.0016, 0.0016, 0.0027, 0.0021, 0.0026], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:55:42,209 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:46,327 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:56:02,446 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.245e+02 2.734e+02 3.789e+02 7.990e+02, threshold=5.469e+02, percent-clipped=5.0 2022-12-08 01:56:05,341 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0188, 3.2629, 4.7259, 3.7098, 4.7402, 4.6260, 4.4428, 4.1490], device='cuda:1'), covar=tensor([0.0488, 0.2775, 0.0827, 0.1511, 0.0669, 0.0658, 0.1640, 0.1684], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0313, 0.0398, 0.0304, 0.0375, 0.0319, 0.0368, 0.0311], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:56:20,043 INFO [train.py:873] (1/4) Epoch 12, batch 3300, loss[loss=0.1378, simple_loss=0.1678, pruned_loss=0.05389, over 14324.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1554, pruned_loss=0.04742, over 2004751.08 frames. ], batch size: 55, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:56:35,433 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86501.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:56:39,486 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86506.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:57:01,791 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86531.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:30,031 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.093e+02 2.507e+02 3.209e+02 6.718e+02, threshold=5.013e+02, percent-clipped=2.0 2022-12-08 01:57:42,891 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86579.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:43,927 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86580.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:46,319 INFO [train.py:873] (1/4) Epoch 12, batch 3400, loss[loss=0.1024, simple_loss=0.1497, pruned_loss=0.02755, over 14673.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1546, pruned_loss=0.04649, over 1984732.35 frames. ], batch size: 23, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:57:48,903 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86586.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:58:03,560 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86602.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:13,173 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86613.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:17,581 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0032, 1.9152, 1.7813, 1.9553, 1.8065, 0.9446, 1.6772, 2.0132], device='cuda:1'), covar=tensor([0.0822, 0.0703, 0.1333, 0.1545, 0.1057, 0.0946, 0.1141, 0.0836], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0027, 0.0030, 0.0026, 0.0029, 0.0040, 0.0028, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 01:58:31,348 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86634.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:58:38,030 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:52,754 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86658.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:55,346 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86661.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:57,881 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.267e+02 2.858e+02 3.502e+02 8.320e+02, threshold=5.717e+02, percent-clipped=7.0 2022-12-08 01:58:58,003 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8875, 2.7113, 2.4431, 2.6089, 2.8262, 2.8227, 2.8584, 2.8312], device='cuda:1'), covar=tensor([0.1135, 0.0744, 0.2324, 0.2447, 0.0949, 0.1099, 0.1240, 0.1050], device='cuda:1'), in_proj_covar=tensor([0.0374, 0.0258, 0.0428, 0.0551, 0.0328, 0.0420, 0.0383, 0.0364], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 01:59:14,482 INFO [train.py:873] (1/4) Epoch 12, batch 3500, loss[loss=0.1156, simple_loss=0.1495, pruned_loss=0.0409, over 14303.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1554, pruned_loss=0.04746, over 1929011.93 frames. ], batch size: 39, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:59:22,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3307, 5.1219, 4.8115, 5.4129, 4.9458, 4.7328, 5.3981, 5.2480], device='cuda:1'), covar=tensor([0.0574, 0.0673, 0.0702, 0.0472, 0.0609, 0.0449, 0.0456, 0.0588], device='cuda:1'), in_proj_covar=tensor([0.0132, 0.0128, 0.0136, 0.0145, 0.0138, 0.0111, 0.0154, 0.0135], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 01:59:34,038 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86706.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:59:35,728 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86708.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:59:51,328 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86725.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:17,807 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86756.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:24,901 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.023e+02 2.697e+02 3.316e+02 6.067e+02, threshold=5.393e+02, percent-clipped=1.0 2022-12-08 02:00:32,573 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86773.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:41,041 INFO [train.py:873] (1/4) Epoch 12, batch 3600, loss[loss=0.1381, simple_loss=0.1393, pruned_loss=0.06843, over 2642.00 frames. ], tot_loss[loss=0.1258, simple_loss=0.1558, pruned_loss=0.0479, over 1904029.76 frames. ], batch size: 100, lr: 6.60e-03, grad_scale: 8.0 2022-12-08 02:00:52,466 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86796.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:57,589 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86801.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:01:08,652 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:01:51,778 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 2.201e+02 2.694e+02 3.354e+02 5.968e+02, threshold=5.388e+02, percent-clipped=2.0 2022-12-08 02:02:01,383 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:04,411 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86878.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:08,035 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7250, 1.7227, 1.7060, 1.7398, 1.6754, 1.1055, 1.4764, 1.6804], device='cuda:1'), covar=tensor([0.1270, 0.0775, 0.1017, 0.0882, 0.1302, 0.0970, 0.0973, 0.1037], device='cuda:1'), in_proj_covar=tensor([0.0027, 0.0027, 0.0030, 0.0026, 0.0029, 0.0040, 0.0028, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 02:02:08,669 INFO [train.py:873] (1/4) Epoch 12, batch 3700, loss[loss=0.1171, simple_loss=0.1531, pruned_loss=0.04054, over 14265.00 frames. ], tot_loss[loss=0.1267, simple_loss=0.1561, pruned_loss=0.04863, over 1882710.24 frames. ], batch size: 57, lr: 6.60e-03, grad_scale: 8.0 2022-12-08 02:02:24,572 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86902.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:25,039 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.25 vs. limit=5.0 2022-12-08 02:02:47,423 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.03 vs. limit=5.0 2022-12-08 02:02:54,823 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:57,465 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:03:06,351 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:03:15,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8055, 1.1868, 2.0093, 1.3103, 1.9077, 2.0485, 1.6585, 2.0544], device='cuda:1'), covar=tensor([0.0368, 0.2037, 0.0553, 0.1855, 0.0700, 0.0677, 0.1172, 0.0485], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0161, 0.0170, 0.0171, 0.0178, 0.0133, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 02:03:19,132 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.304e+02 2.133e+02 2.735e+02 3.535e+02 1.030e+03, threshold=5.470e+02, percent-clipped=7.0 2022-12-08 02:03:26,182 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1812, 1.4715, 4.1065, 1.7587, 4.0098, 4.1260, 3.3541, 4.5094], device='cuda:1'), covar=tensor([0.0223, 0.3249, 0.0395, 0.2327, 0.0366, 0.0366, 0.0534, 0.0149], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0161, 0.0170, 0.0171, 0.0178, 0.0134, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:03:35,277 INFO [train.py:873] (1/4) Epoch 12, batch 3800, loss[loss=0.1089, simple_loss=0.1497, pruned_loss=0.03401, over 14566.00 frames. ], tot_loss[loss=0.1264, simple_loss=0.1561, pruned_loss=0.04834, over 1925551.48 frames. ], batch size: 34, lr: 6.60e-03, grad_scale: 4.0 2022-12-08 02:03:35,727 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-08 02:03:39,304 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 02:03:52,821 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87002.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:04:13,627 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 02:04:45,849 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87063.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:04:47,286 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.236e+02 2.550e+02 3.017e+02 4.626e+02, threshold=5.100e+02, percent-clipped=0.0 2022-12-08 02:05:03,579 INFO [train.py:873] (1/4) Epoch 12, batch 3900, loss[loss=0.1099, simple_loss=0.1447, pruned_loss=0.03755, over 14581.00 frames. ], tot_loss[loss=0.1255, simple_loss=0.1554, pruned_loss=0.04786, over 1887177.37 frames. ], batch size: 43, lr: 6.59e-03, grad_scale: 4.0 2022-12-08 02:05:07,581 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 02:05:14,059 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87096.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:05:18,332 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87101.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:05:49,272 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2022-12-08 02:05:55,275 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1117, 1.9788, 3.2441, 2.3606, 3.0879, 1.9061, 2.5213, 3.0863], device='cuda:1'), covar=tensor([0.0857, 0.3844, 0.0485, 0.5549, 0.0750, 0.3177, 0.1358, 0.0601], device='cuda:1'), in_proj_covar=tensor([0.0245, 0.0210, 0.0202, 0.0286, 0.0222, 0.0214, 0.0212, 0.0206], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:05:56,022 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87144.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:00,194 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87149.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:14,161 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.280e+02 2.775e+02 3.261e+02 6.407e+02, threshold=5.550e+02, percent-clipped=4.0 2022-12-08 02:06:18,648 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:23,488 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-08 02:06:29,720 INFO [train.py:873] (1/4) Epoch 12, batch 4000, loss[loss=0.1908, simple_loss=0.1696, pruned_loss=0.106, over 1233.00 frames. ], tot_loss[loss=0.125, simple_loss=0.155, pruned_loss=0.04753, over 1901331.68 frames. ], batch size: 100, lr: 6.59e-03, grad_scale: 8.0 2022-12-08 02:07:10,572 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87230.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:14,081 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:15,971 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87236.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:40,897 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.153e+02 2.853e+02 3.504e+02 7.130e+02, threshold=5.705e+02, percent-clipped=5.0 2022-12-08 02:07:57,456 INFO [train.py:873] (1/4) Epoch 12, batch 4100, loss[loss=0.1568, simple_loss=0.1769, pruned_loss=0.06836, over 7770.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1544, pruned_loss=0.04655, over 1929117.67 frames. ], batch size: 100, lr: 6.58e-03, grad_scale: 8.0 2022-12-08 02:07:58,294 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:08:04,404 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87291.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:08:14,536 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1912, 1.2578, 1.4169, 1.0797, 0.8564, 1.2389, 0.8125, 1.3015], device='cuda:1'), covar=tensor([0.1752, 0.3105, 0.1028, 0.2523, 0.3594, 0.1108, 0.2398, 0.1114], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0097, 0.0091, 0.0096, 0.0115, 0.0085, 0.0123, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:08:17,150 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3253, 2.6417, 2.6435, 2.7402, 2.1913, 2.6994, 2.5533, 1.3945], device='cuda:1'), covar=tensor([0.1974, 0.0904, 0.0777, 0.0522, 0.1141, 0.0754, 0.1229, 0.2742], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0078, 0.0063, 0.0067, 0.0091, 0.0077, 0.0094, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 02:09:02,748 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87358.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:09:06,567 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87362.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:09:09,785 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.165e+02 2.894e+02 3.729e+02 1.916e+03, threshold=5.789e+02, percent-clipped=7.0 2022-12-08 02:09:24,060 INFO [train.py:873] (1/4) Epoch 12, batch 4200, loss[loss=0.1309, simple_loss=0.1657, pruned_loss=0.04809, over 14303.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1546, pruned_loss=0.04642, over 1946228.77 frames. ], batch size: 76, lr: 6.58e-03, grad_scale: 4.0 2022-12-08 02:09:50,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3870, 4.8554, 4.8509, 5.3749, 5.0059, 4.4371, 5.2850, 4.4401], device='cuda:1'), covar=tensor([0.0342, 0.1101, 0.0356, 0.0388, 0.0807, 0.0538, 0.0529, 0.0516], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0268, 0.0189, 0.0187, 0.0180, 0.0155, 0.0275, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 02:09:58,954 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87423.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:19,626 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0548, 2.0206, 2.3156, 1.5863, 1.6859, 2.1582, 1.2829, 2.1059], device='cuda:1'), covar=tensor([0.1164, 0.1597, 0.0820, 0.2424, 0.2513, 0.0909, 0.3614, 0.1213], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0097, 0.0091, 0.0097, 0.0115, 0.0085, 0.0123, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:10:24,657 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9623, 1.7634, 1.5371, 1.9140, 1.8985, 1.6221, 1.5662, 1.3632], device='cuda:1'), covar=tensor([0.0318, 0.0732, 0.0624, 0.0745, 0.0420, 0.0602, 0.0517, 0.1233], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0017, 0.0015, 0.0016, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 02:10:32,222 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87461.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:36,120 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.802e+01 2.101e+02 2.786e+02 3.317e+02 5.801e+02, threshold=5.572e+02, percent-clipped=2.0 2022-12-08 02:10:39,562 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87470.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:42,120 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3803, 2.1106, 2.3402, 1.6222, 1.9882, 2.3984, 2.4748, 2.0206], device='cuda:1'), covar=tensor([0.0730, 0.0828, 0.0851, 0.1385, 0.1121, 0.0701, 0.0531, 0.1424], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0180, 0.0138, 0.0125, 0.0135, 0.0144, 0.0122, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 02:10:51,434 INFO [train.py:873] (1/4) Epoch 12, batch 4300, loss[loss=0.1075, simple_loss=0.1482, pruned_loss=0.03339, over 14550.00 frames. ], tot_loss[loss=0.1247, simple_loss=0.1556, pruned_loss=0.0469, over 2007986.31 frames. ], batch size: 43, lr: 6.58e-03, grad_scale: 4.0 2022-12-08 02:10:55,836 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-08 02:11:10,036 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 02:11:10,417 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:21,395 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87518.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:25,983 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87522.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:35,598 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1073, 2.2896, 4.0367, 4.1535, 4.0408, 2.4474, 4.0643, 3.0826], device='cuda:1'), covar=tensor([0.0370, 0.1034, 0.0773, 0.0415, 0.0361, 0.1520, 0.0364, 0.0952], device='cuda:1'), in_proj_covar=tensor([0.0288, 0.0256, 0.0375, 0.0326, 0.0267, 0.0304, 0.0305, 0.0284], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 02:11:36,408 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87534.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:04,323 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.182e+02 2.709e+02 3.348e+02 1.136e+03, threshold=5.418e+02, percent-clipped=3.0 2022-12-08 02:12:04,556 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87566.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:12:18,397 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87582.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:19,183 INFO [train.py:873] (1/4) Epoch 12, batch 4400, loss[loss=0.1073, simple_loss=0.1523, pruned_loss=0.03109, over 14213.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1554, pruned_loss=0.04674, over 1991608.60 frames. ], batch size: 32, lr: 6.57e-03, grad_scale: 8.0 2022-12-08 02:12:19,909 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87584.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:21,531 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87586.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:24,085 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87589.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:38,840 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:13:01,920 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2661, 4.9834, 4.5518, 4.7400, 4.8271, 5.0828, 5.2140, 5.1878], device='cuda:1'), covar=tensor([0.0620, 0.0430, 0.2205, 0.2756, 0.0705, 0.0802, 0.0730, 0.0736], device='cuda:1'), in_proj_covar=tensor([0.0375, 0.0259, 0.0436, 0.0556, 0.0330, 0.0423, 0.0387, 0.0363], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:13:13,631 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87645.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:13:18,241 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:13:24,905 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87658.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:13:25,263 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 02:13:31,881 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87666.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:13:32,488 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.847e+01 2.130e+02 2.616e+02 3.250e+02 5.248e+02, threshold=5.232e+02, percent-clipped=0.0 2022-12-08 02:13:46,587 INFO [train.py:873] (1/4) Epoch 12, batch 4500, loss[loss=0.1108, simple_loss=0.153, pruned_loss=0.03428, over 13888.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1548, pruned_loss=0.04628, over 1936306.45 frames. ], batch size: 20, lr: 6.57e-03, grad_scale: 4.0 2022-12-08 02:14:06,777 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87706.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:14:17,088 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:14:30,644 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2022-12-08 02:14:55,998 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3409, 2.3475, 2.4168, 2.4878, 2.4201, 2.1507, 1.3921, 2.1460], device='cuda:1'), covar=tensor([0.0627, 0.0635, 0.0647, 0.0378, 0.0654, 0.1089, 0.2645, 0.0568], device='cuda:1'), in_proj_covar=tensor([0.0159, 0.0171, 0.0141, 0.0138, 0.0199, 0.0138, 0.0157, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 02:14:59,665 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.389e+02 2.965e+02 3.391e+02 5.507e+02, threshold=5.931e+02, percent-clipped=1.0 2022-12-08 02:15:13,018 INFO [train.py:873] (1/4) Epoch 12, batch 4600, loss[loss=0.1487, simple_loss=0.1785, pruned_loss=0.05942, over 14636.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1558, pruned_loss=0.04719, over 1938897.37 frames. ], batch size: 22, lr: 6.57e-03, grad_scale: 4.0 2022-12-08 02:15:16,089 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2022-12-08 02:15:39,139 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0631, 4.6627, 4.5323, 5.0922, 4.6959, 4.3527, 4.9987, 4.2105], device='cuda:1'), covar=tensor([0.0326, 0.1086, 0.0350, 0.0350, 0.0819, 0.0563, 0.0535, 0.0468], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0264, 0.0188, 0.0183, 0.0178, 0.0151, 0.0272, 0.0163], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 02:15:43,350 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87817.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:15:55,614 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-08 02:16:21,503 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87861.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:16:26,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.249e+02 2.775e+02 3.473e+02 5.522e+02, threshold=5.551e+02, percent-clipped=0.0 2022-12-08 02:16:37,958 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4145, 3.3063, 3.2437, 3.4754, 3.1202, 2.8620, 3.4776, 3.3896], device='cuda:1'), covar=tensor([0.0692, 0.1010, 0.0794, 0.0654, 0.1039, 0.0861, 0.0649, 0.0713], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0129, 0.0135, 0.0146, 0.0137, 0.0112, 0.0156, 0.0134], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:16:40,648 INFO [train.py:873] (1/4) Epoch 12, batch 4700, loss[loss=0.1245, simple_loss=0.1529, pruned_loss=0.04806, over 6964.00 frames. ], tot_loss[loss=0.125, simple_loss=0.1557, pruned_loss=0.04718, over 1936404.63 frames. ], batch size: 100, lr: 6.56e-03, grad_scale: 4.0 2022-12-08 02:16:43,358 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87886.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:02,652 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-08 02:17:25,802 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87934.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:30,713 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 02:17:31,138 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87940.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:17:34,571 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9660, 2.5996, 3.3705, 2.1151, 2.1330, 2.9856, 1.6518, 2.8324], device='cuda:1'), covar=tensor([0.0833, 0.1276, 0.0718, 0.2332, 0.2421, 0.0816, 0.3505, 0.0974], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0097, 0.0091, 0.0096, 0.0114, 0.0085, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:17:35,270 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:48,987 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87961.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:17:54,230 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.984e+01 2.201e+02 2.736e+02 3.426e+02 5.956e+02, threshold=5.473e+02, percent-clipped=2.0 2022-12-08 02:17:57,034 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 02:18:08,684 INFO [train.py:873] (1/4) Epoch 12, batch 4800, loss[loss=0.1546, simple_loss=0.1456, pruned_loss=0.08179, over 2622.00 frames. ], tot_loss[loss=0.125, simple_loss=0.1552, pruned_loss=0.04743, over 1894822.76 frames. ], batch size: 100, lr: 6.56e-03, grad_scale: 8.0 2022-12-08 02:18:39,753 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:19:14,900 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9030, 1.5070, 3.1374, 2.8657, 3.0310, 3.1058, 2.2997, 3.1444], device='cuda:1'), covar=tensor([0.1057, 0.1232, 0.0125, 0.0292, 0.0257, 0.0137, 0.0439, 0.0143], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0128, 0.0166, 0.0144, 0.0141, 0.0120, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 02:19:21,830 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:19:22,466 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.264e+02 2.796e+02 3.655e+02 7.619e+02, threshold=5.592e+02, percent-clipped=2.0 2022-12-08 02:19:36,608 INFO [train.py:873] (1/4) Epoch 12, batch 4900, loss[loss=0.1349, simple_loss=0.1603, pruned_loss=0.05475, over 6926.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1546, pruned_loss=0.0464, over 1953769.72 frames. ], batch size: 100, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:19:48,800 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7603, 2.5065, 3.1505, 1.9665, 1.9858, 2.8462, 1.4471, 2.6900], device='cuda:1'), covar=tensor([0.0745, 0.1128, 0.0633, 0.2779, 0.2353, 0.0736, 0.3419, 0.0836], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0097, 0.0091, 0.0097, 0.0115, 0.0085, 0.0123, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:19:51,943 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88100.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:06,808 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:26,706 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1215, 3.8692, 3.7511, 4.1902, 3.9178, 3.6250, 4.1714, 3.4938], device='cuda:1'), covar=tensor([0.0600, 0.0938, 0.0446, 0.0437, 0.0770, 0.1307, 0.0563, 0.0531], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0264, 0.0187, 0.0184, 0.0176, 0.0151, 0.0272, 0.0163], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 02:20:39,830 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.47 vs. limit=2.0 2022-12-08 02:20:45,894 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88161.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:20:45,944 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88161.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:49,494 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:51,095 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.168e+02 2.607e+02 3.248e+02 6.908e+02, threshold=5.213e+02, percent-clipped=2.0 2022-12-08 02:21:05,439 INFO [train.py:873] (1/4) Epoch 12, batch 5000, loss[loss=0.1355, simple_loss=0.1594, pruned_loss=0.05579, over 13521.00 frames. ], tot_loss[loss=0.1249, simple_loss=0.1552, pruned_loss=0.04732, over 1906058.25 frames. ], batch size: 100, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:21:28,452 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88209.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:21:55,777 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88240.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:00,486 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88245.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:14,901 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88261.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:22:19,850 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.015e+02 2.163e+02 2.571e+02 3.278e+02 8.920e+02, threshold=5.143e+02, percent-clipped=2.0 2022-12-08 02:22:33,614 INFO [train.py:873] (1/4) Epoch 12, batch 5100, loss[loss=0.1363, simple_loss=0.1625, pruned_loss=0.05503, over 14220.00 frames. ], tot_loss[loss=0.1243, simple_loss=0.1549, pruned_loss=0.04685, over 1942190.76 frames. ], batch size: 60, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:22:38,215 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88288.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:41,230 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88291.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:22:42,680 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:56,553 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88309.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:23:34,059 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88352.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:23:46,914 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.245e+02 2.755e+02 3.354e+02 6.298e+02, threshold=5.511e+02, percent-clipped=3.0 2022-12-08 02:24:01,890 INFO [train.py:873] (1/4) Epoch 12, batch 5200, loss[loss=0.1162, simple_loss=0.1561, pruned_loss=0.03815, over 14301.00 frames. ], tot_loss[loss=0.1239, simple_loss=0.1546, pruned_loss=0.04662, over 1946608.52 frames. ], batch size: 60, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:25:06,279 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88456.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:25:15,464 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.113e+02 2.685e+02 3.247e+02 4.954e+02, threshold=5.369e+02, percent-clipped=0.0 2022-12-08 02:25:29,579 INFO [train.py:873] (1/4) Epoch 12, batch 5300, loss[loss=0.1918, simple_loss=0.1642, pruned_loss=0.1097, over 1258.00 frames. ], tot_loss[loss=0.123, simple_loss=0.1539, pruned_loss=0.04602, over 1919348.11 frames. ], batch size: 100, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:26:13,202 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8185, 1.3107, 2.7296, 2.4533, 2.6589, 2.7036, 1.8303, 2.7292], device='cuda:1'), covar=tensor([0.1211, 0.1484, 0.0159, 0.0402, 0.0404, 0.0172, 0.0567, 0.0224], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0128, 0.0168, 0.0146, 0.0142, 0.0121, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 02:26:43,274 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.234e+01 2.154e+02 2.570e+02 3.365e+02 7.012e+02, threshold=5.140e+02, percent-clipped=5.0 2022-12-08 02:26:57,898 INFO [train.py:873] (1/4) Epoch 12, batch 5400, loss[loss=0.1207, simple_loss=0.1566, pruned_loss=0.04236, over 14269.00 frames. ], tot_loss[loss=0.1234, simple_loss=0.1543, pruned_loss=0.04619, over 1920241.04 frames. ], batch size: 69, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:27:03,436 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5273, 3.1003, 2.4379, 3.5998, 3.4883, 3.4150, 3.0183, 2.3346], device='cuda:1'), covar=tensor([0.0681, 0.1506, 0.3501, 0.0623, 0.0825, 0.1289, 0.1312, 0.3642], device='cuda:1'), in_proj_covar=tensor([0.0262, 0.0293, 0.0263, 0.0261, 0.0306, 0.0291, 0.0252, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:1') 2022-12-08 02:27:12,001 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0333, 2.2241, 4.1025, 2.8295, 3.8848, 1.9725, 3.1003, 3.8495], device='cuda:1'), covar=tensor([0.0709, 0.3858, 0.0408, 0.6207, 0.0645, 0.3284, 0.1358, 0.0498], device='cuda:1'), in_proj_covar=tensor([0.0246, 0.0209, 0.0203, 0.0281, 0.0223, 0.0211, 0.0211, 0.0207], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:27:46,064 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3643, 3.4181, 4.1639, 2.9372, 2.4669, 3.6028, 1.9260, 3.4827], device='cuda:1'), covar=tensor([0.1131, 0.1021, 0.0837, 0.2479, 0.2104, 0.0885, 0.3198, 0.1468], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0097, 0.0091, 0.0097, 0.0116, 0.0084, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:27:48,546 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7127, 5.1901, 5.2123, 5.7205, 5.2565, 4.6329, 5.6794, 4.5605], device='cuda:1'), covar=tensor([0.0290, 0.0969, 0.0281, 0.0348, 0.0743, 0.0460, 0.0467, 0.0547], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0262, 0.0186, 0.0182, 0.0177, 0.0150, 0.0271, 0.0161], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 02:27:54,012 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88647.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:28:11,935 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.088e+01 2.108e+02 2.673e+02 3.283e+02 5.238e+02, threshold=5.346e+02, percent-clipped=1.0 2022-12-08 02:28:17,390 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4894, 2.3896, 2.7712, 1.7822, 1.9951, 2.5447, 1.5477, 2.4515], device='cuda:1'), covar=tensor([0.0937, 0.1565, 0.0696, 0.3131, 0.2418, 0.0970, 0.3566, 0.1051], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0097, 0.0091, 0.0097, 0.0116, 0.0084, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:28:26,023 INFO [train.py:873] (1/4) Epoch 12, batch 5500, loss[loss=0.1197, simple_loss=0.154, pruned_loss=0.04273, over 13938.00 frames. ], tot_loss[loss=0.1232, simple_loss=0.154, pruned_loss=0.04623, over 1963848.94 frames. ], batch size: 20, lr: 6.53e-03, grad_scale: 8.0 2022-12-08 02:28:28,037 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1802, 2.8065, 2.8351, 1.9953, 2.5817, 2.8382, 3.1147, 2.4526], device='cuda:1'), covar=tensor([0.0650, 0.1083, 0.0938, 0.1621, 0.1122, 0.0772, 0.0868, 0.1505], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0180, 0.0138, 0.0126, 0.0136, 0.0145, 0.0123, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 02:28:45,554 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7410, 2.4660, 3.3171, 1.9625, 2.1549, 3.0251, 1.5674, 2.8761], device='cuda:1'), covar=tensor([0.0994, 0.1435, 0.0581, 0.3304, 0.2259, 0.0745, 0.3603, 0.0902], device='cuda:1'), in_proj_covar=tensor([0.0080, 0.0097, 0.0090, 0.0097, 0.0115, 0.0084, 0.0121, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:29:29,738 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88756.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:29:38,829 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.096e+02 2.766e+02 3.641e+02 8.752e+02, threshold=5.532e+02, percent-clipped=4.0 2022-12-08 02:29:53,655 INFO [train.py:873] (1/4) Epoch 12, batch 5600, loss[loss=0.132, simple_loss=0.1541, pruned_loss=0.05497, over 6023.00 frames. ], tot_loss[loss=0.1236, simple_loss=0.1548, pruned_loss=0.04623, over 2004251.54 frames. ], batch size: 100, lr: 6.53e-03, grad_scale: 8.0 2022-12-08 02:30:12,306 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88804.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:30:34,303 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=8.09 vs. limit=5.0 2022-12-08 02:31:06,998 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.217e+02 2.679e+02 3.309e+02 6.646e+02, threshold=5.358e+02, percent-clipped=2.0 2022-12-08 02:31:18,261 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6694, 4.3973, 4.0895, 4.2504, 4.4102, 4.5204, 4.6305, 4.5797], device='cuda:1'), covar=tensor([0.0730, 0.0463, 0.1979, 0.2487, 0.0731, 0.0722, 0.0727, 0.0808], device='cuda:1'), in_proj_covar=tensor([0.0377, 0.0261, 0.0434, 0.0555, 0.0327, 0.0426, 0.0385, 0.0366], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:31:20,847 INFO [train.py:873] (1/4) Epoch 12, batch 5700, loss[loss=0.1226, simple_loss=0.1569, pruned_loss=0.04414, over 6959.00 frames. ], tot_loss[loss=0.125, simple_loss=0.1558, pruned_loss=0.04709, over 1998991.66 frames. ], batch size: 100, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:32:16,448 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0638, 3.8359, 3.7537, 4.1074, 3.7081, 3.3419, 4.1353, 4.0160], device='cuda:1'), covar=tensor([0.0626, 0.0807, 0.0818, 0.0605, 0.0817, 0.0942, 0.0616, 0.0741], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0131, 0.0140, 0.0149, 0.0138, 0.0116, 0.0158, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:32:16,523 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88947.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:32:32,856 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 2.272e+02 2.731e+02 3.384e+02 1.131e+03, threshold=5.463e+02, percent-clipped=4.0 2022-12-08 02:32:47,396 INFO [train.py:873] (1/4) Epoch 12, batch 5800, loss[loss=0.1144, simple_loss=0.1545, pruned_loss=0.0372, over 14098.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1555, pruned_loss=0.04675, over 2020683.70 frames. ], batch size: 29, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:32:50,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5748, 2.4604, 4.3635, 4.5774, 4.3826, 2.4793, 4.4943, 3.4893], device='cuda:1'), covar=tensor([0.0326, 0.1035, 0.0686, 0.0324, 0.0331, 0.1558, 0.0315, 0.0886], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0256, 0.0374, 0.0325, 0.0266, 0.0302, 0.0304, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 02:32:57,689 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88995.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:33:26,290 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:34:01,515 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.293e+02 2.084e+02 2.490e+02 3.041e+02 5.969e+02, threshold=4.980e+02, percent-clipped=1.0 2022-12-08 02:34:15,417 INFO [train.py:873] (1/4) Epoch 12, batch 5900, loss[loss=0.1279, simple_loss=0.1584, pruned_loss=0.04873, over 14469.00 frames. ], tot_loss[loss=0.1249, simple_loss=0.1554, pruned_loss=0.04722, over 1971162.19 frames. ], batch size: 49, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:34:20,398 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:35:01,842 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1877, 2.1736, 3.1142, 3.2321, 3.1340, 2.2130, 3.1021, 2.4725], device='cuda:1'), covar=tensor([0.0381, 0.0913, 0.0753, 0.0412, 0.0453, 0.1241, 0.0363, 0.0856], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0256, 0.0371, 0.0326, 0.0267, 0.0303, 0.0304, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 02:35:29,920 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.297e+02 2.664e+02 3.393e+02 7.245e+02, threshold=5.328e+02, percent-clipped=3.0 2022-12-08 02:35:31,824 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89169.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:35:44,587 INFO [train.py:873] (1/4) Epoch 12, batch 6000, loss[loss=0.1354, simple_loss=0.1639, pruned_loss=0.0534, over 13545.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1547, pruned_loss=0.04652, over 1960900.30 frames. ], batch size: 100, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:35:44,587 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 02:35:52,918 INFO [train.py:905] (1/4) Epoch 12, validation: loss=0.1296, simple_loss=0.1695, pruned_loss=0.04492, over 857387.00 frames. 2022-12-08 02:35:52,919 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 02:36:34,395 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89230.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:37:06,542 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.059e+02 2.275e+02 2.670e+02 3.684e+02 9.566e+02, threshold=5.341e+02, percent-clipped=7.0 2022-12-08 02:37:10,218 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.94 vs. limit=5.0 2022-12-08 02:37:21,419 INFO [train.py:873] (1/4) Epoch 12, batch 6100, loss[loss=0.1582, simple_loss=0.1796, pruned_loss=0.06844, over 13530.00 frames. ], tot_loss[loss=0.1249, simple_loss=0.1555, pruned_loss=0.04716, over 2003581.80 frames. ], batch size: 100, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:38:34,354 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.240e+02 2.644e+02 3.331e+02 6.090e+02, threshold=5.289e+02, percent-clipped=8.0 2022-12-08 02:38:40,666 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:38:48,056 INFO [train.py:873] (1/4) Epoch 12, batch 6200, loss[loss=0.1485, simple_loss=0.1615, pruned_loss=0.06769, over 6947.00 frames. ], tot_loss[loss=0.1257, simple_loss=0.1557, pruned_loss=0.04782, over 1932997.58 frames. ], batch size: 100, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:38:48,164 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:39:24,568 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:39:33,931 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:39:37,264 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89439.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:40:00,789 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.324e+02 2.890e+02 3.661e+02 8.443e+02, threshold=5.779e+02, percent-clipped=4.0 2022-12-08 02:40:15,101 INFO [train.py:873] (1/4) Epoch 12, batch 6300, loss[loss=0.1072, simple_loss=0.1491, pruned_loss=0.03268, over 14222.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1549, pruned_loss=0.04635, over 1954753.76 frames. ], batch size: 37, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:40:17,846 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:40:30,002 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89500.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:40:37,606 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89509.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:40:37,671 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2311, 2.7832, 4.0348, 3.2627, 4.1035, 3.8728, 3.8661, 3.4027], device='cuda:1'), covar=tensor([0.0753, 0.2808, 0.0903, 0.1607, 0.0743, 0.0896, 0.1525, 0.1946], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0317, 0.0397, 0.0305, 0.0378, 0.0321, 0.0367, 0.0309], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:40:51,646 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89525.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:41:24,175 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2022-12-08 02:41:27,938 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.271e+01 2.078e+02 2.565e+02 3.125e+02 6.537e+02, threshold=5.131e+02, percent-clipped=2.0 2022-12-08 02:41:30,766 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89570.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:41:41,866 INFO [train.py:873] (1/4) Epoch 12, batch 6400, loss[loss=0.1434, simple_loss=0.1473, pruned_loss=0.06973, over 3870.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1549, pruned_loss=0.04635, over 1953187.36 frames. ], batch size: 100, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:41:57,507 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0564, 2.1395, 3.1299, 3.2282, 3.1249, 2.0991, 3.1730, 2.3096], device='cuda:1'), covar=tensor([0.0430, 0.1007, 0.0680, 0.0469, 0.0436, 0.1444, 0.0387, 0.1039], device='cuda:1'), in_proj_covar=tensor([0.0288, 0.0253, 0.0369, 0.0323, 0.0264, 0.0301, 0.0303, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:42:34,505 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5348, 2.4021, 2.9753, 2.0130, 2.0225, 2.6077, 1.4538, 2.5661], device='cuda:1'), covar=tensor([0.0999, 0.1218, 0.0811, 0.2159, 0.2632, 0.0903, 0.4064, 0.1149], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0098, 0.0092, 0.0097, 0.0116, 0.0085, 0.0123, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:42:48,646 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7889, 1.4856, 1.7837, 2.0615, 1.3529, 1.7891, 1.6484, 1.9014], device='cuda:1'), covar=tensor([0.0143, 0.0231, 0.0123, 0.0122, 0.0243, 0.0256, 0.0165, 0.0102], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0253, 0.0369, 0.0322, 0.0263, 0.0301, 0.0302, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:42:56,147 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.320e+02 2.969e+02 3.886e+02 7.379e+02, threshold=5.938e+02, percent-clipped=5.0 2022-12-08 02:43:07,297 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:09,689 INFO [train.py:873] (1/4) Epoch 12, batch 6500, loss[loss=0.1359, simple_loss=0.149, pruned_loss=0.06142, over 3865.00 frames. ], tot_loss[loss=0.1242, simple_loss=0.1552, pruned_loss=0.04658, over 1960769.41 frames. ], batch size: 100, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:43:09,846 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89683.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:50,393 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89730.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:51,208 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:00,175 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89741.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:23,988 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.343e+02 2.899e+02 3.659e+02 1.529e+03, threshold=5.799e+02, percent-clipped=6.0 2022-12-08 02:44:24,569 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.17 vs. limit=5.0 2022-12-08 02:44:35,254 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:36,899 INFO [train.py:873] (1/4) Epoch 12, batch 6600, loss[loss=0.1253, simple_loss=0.1593, pruned_loss=0.04564, over 14375.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1554, pruned_loss=0.04677, over 2004111.24 frames. ], batch size: 73, lr: 6.49e-03, grad_scale: 8.0 2022-12-08 02:44:47,385 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89795.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:45:12,087 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4228, 2.3979, 1.9879, 2.4630, 2.2813, 2.3688, 2.2340, 2.0152], device='cuda:1'), covar=tensor([0.0725, 0.0912, 0.2074, 0.0798, 0.1017, 0.0813, 0.1306, 0.1325], device='cuda:1'), in_proj_covar=tensor([0.0267, 0.0297, 0.0268, 0.0263, 0.0310, 0.0293, 0.0256, 0.0248], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:45:13,787 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89825.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:48,195 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:48,289 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:50,623 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.462e+01 2.267e+02 2.755e+02 3.423e+02 7.380e+02, threshold=5.509e+02, percent-clipped=4.0 2022-12-08 02:45:55,151 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89873.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:46:02,695 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8568, 5.5061, 5.3497, 5.8701, 5.3533, 5.1419, 5.9187, 5.7057], device='cuda:1'), covar=tensor([0.0484, 0.0567, 0.0689, 0.0411, 0.0660, 0.0339, 0.0470, 0.0527], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0132, 0.0139, 0.0149, 0.0139, 0.0115, 0.0158, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:46:04,380 INFO [train.py:873] (1/4) Epoch 12, batch 6700, loss[loss=0.134, simple_loss=0.1464, pruned_loss=0.06083, over 3754.00 frames. ], tot_loss[loss=0.1246, simple_loss=0.1555, pruned_loss=0.0469, over 1983865.27 frames. ], batch size: 100, lr: 6.49e-03, grad_scale: 8.0 2022-12-08 02:46:06,468 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 02:46:41,514 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89926.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:47:02,787 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8886, 3.6489, 3.3014, 2.6501, 3.3183, 3.5789, 4.0009, 3.1128], device='cuda:1'), covar=tensor([0.0498, 0.1456, 0.0872, 0.1303, 0.0805, 0.0563, 0.0551, 0.1091], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0176, 0.0135, 0.0123, 0.0133, 0.0145, 0.0122, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 02:47:17,826 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.198e+02 2.718e+02 3.192e+02 1.007e+03, threshold=5.437e+02, percent-clipped=1.0 2022-12-08 02:47:30,899 INFO [train.py:873] (1/4) Epoch 12, batch 6800, loss[loss=0.1131, simple_loss=0.1497, pruned_loss=0.03822, over 13949.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1552, pruned_loss=0.04685, over 1942110.26 frames. ], batch size: 23, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:48:14,837 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2022-12-08 02:48:15,286 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90030.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:20,544 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:33,132 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3307, 3.6012, 3.5303, 3.5226, 2.7158, 3.6229, 3.5597, 1.7767], device='cuda:1'), covar=tensor([0.1897, 0.1194, 0.0882, 0.0538, 0.0857, 0.0771, 0.1200, 0.2175], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0079, 0.0064, 0.0067, 0.0093, 0.0079, 0.0095, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 02:48:47,746 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.301e+02 2.958e+02 4.111e+02 8.301e+02, threshold=5.916e+02, percent-clipped=6.0 2022-12-08 02:48:57,166 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:59,929 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90081.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:01,447 INFO [train.py:873] (1/4) Epoch 12, batch 6900, loss[loss=0.1459, simple_loss=0.1495, pruned_loss=0.07117, over 1317.00 frames. ], tot_loss[loss=0.1252, simple_loss=0.1554, pruned_loss=0.04751, over 1925341.81 frames. ], batch size: 100, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:49:11,902 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90095.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:49:41,789 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90129.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:42,830 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:53,769 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90143.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:50:08,683 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 02:50:13,203 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:50:15,735 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.354e+02 2.821e+02 3.688e+02 2.263e+03, threshold=5.641e+02, percent-clipped=6.0 2022-12-08 02:50:28,848 INFO [train.py:873] (1/4) Epoch 12, batch 7000, loss[loss=0.1126, simple_loss=0.1535, pruned_loss=0.03582, over 14207.00 frames. ], tot_loss[loss=0.1256, simple_loss=0.1558, pruned_loss=0.04766, over 1970135.87 frames. ], batch size: 25, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:50:36,034 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90191.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:50:38,290 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 02:50:55,443 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:51:02,619 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90221.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:51:11,995 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90232.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:51:21,618 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3869, 3.1358, 4.0274, 2.8355, 2.3762, 3.6432, 1.9261, 3.4171], device='cuda:1'), covar=tensor([0.1184, 0.1071, 0.0491, 0.2524, 0.2239, 0.0817, 0.3304, 0.1175], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0098, 0.0091, 0.0096, 0.0115, 0.0084, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:51:25,492 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.75 vs. limit=5.0 2022-12-08 02:51:44,201 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 1.997e+02 2.595e+02 3.353e+02 6.185e+02, threshold=5.189e+02, percent-clipped=2.0 2022-12-08 02:51:57,578 INFO [train.py:873] (1/4) Epoch 12, batch 7100, loss[loss=0.1127, simple_loss=0.148, pruned_loss=0.03863, over 14239.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1556, pruned_loss=0.04728, over 1989400.05 frames. ], batch size: 69, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:52:06,207 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:52:06,419 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-08 02:52:37,474 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.70 vs. limit=5.0 2022-12-08 02:52:44,223 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:52:57,297 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90351.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:53:12,653 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 2.346e+02 2.734e+02 3.324e+02 7.118e+02, threshold=5.469e+02, percent-clipped=5.0 2022-12-08 02:53:26,144 INFO [train.py:873] (1/4) Epoch 12, batch 7200, loss[loss=0.09299, simple_loss=0.1363, pruned_loss=0.02486, over 14634.00 frames. ], tot_loss[loss=0.1252, simple_loss=0.1555, pruned_loss=0.04739, over 2001136.69 frames. ], batch size: 22, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:53:26,984 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:53:52,001 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90412.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:54:25,072 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-08 02:54:32,994 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90459.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:54:40,954 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.297e+02 2.801e+02 3.599e+02 6.534e+02, threshold=5.602e+02, percent-clipped=3.0 2022-12-08 02:54:50,150 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1094, 1.3207, 1.3752, 0.9537, 0.9157, 1.1002, 0.8104, 1.2394], device='cuda:1'), covar=tensor([0.1778, 0.2389, 0.0956, 0.2484, 0.3158, 0.1056, 0.2010, 0.1063], device='cuda:1'), in_proj_covar=tensor([0.0081, 0.0097, 0.0090, 0.0096, 0.0114, 0.0084, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 02:54:54,141 INFO [train.py:873] (1/4) Epoch 12, batch 7300, loss[loss=0.1281, simple_loss=0.1535, pruned_loss=0.05141, over 6908.00 frames. ], tot_loss[loss=0.1231, simple_loss=0.1541, pruned_loss=0.04608, over 2023027.61 frames. ], batch size: 100, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:54:56,690 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:05,305 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90496.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:19,964 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9261, 1.8854, 1.6667, 1.9337, 1.8706, 1.8828, 1.7758, 1.7235], device='cuda:1'), covar=tensor([0.0919, 0.0860, 0.1898, 0.0598, 0.0926, 0.0523, 0.1531, 0.0896], device='cuda:1'), in_proj_covar=tensor([0.0271, 0.0299, 0.0271, 0.0265, 0.0314, 0.0298, 0.0258, 0.0249], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:55:26,535 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:27,584 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90521.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:55:58,937 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90557.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:56:04,269 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4579, 3.5311, 3.5651, 3.2809, 3.4960, 3.5318, 1.3990, 3.3422], device='cuda:1'), covar=tensor([0.0470, 0.0662, 0.0682, 0.0847, 0.0637, 0.0591, 0.4178, 0.0491], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0170, 0.0144, 0.0141, 0.0202, 0.0136, 0.0158, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 02:56:08,321 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.439e+02 2.820e+02 3.761e+02 1.640e+03, threshold=5.641e+02, percent-clipped=6.0 2022-12-08 02:56:09,282 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90569.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:56:17,613 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 02:56:21,034 INFO [train.py:873] (1/4) Epoch 12, batch 7400, loss[loss=0.1324, simple_loss=0.1468, pruned_loss=0.05899, over 3830.00 frames. ], tot_loss[loss=0.123, simple_loss=0.1543, pruned_loss=0.04584, over 2039153.23 frames. ], batch size: 100, lr: 6.46e-03, grad_scale: 8.0 2022-12-08 02:56:23,943 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5623, 2.4594, 2.2287, 2.2841, 2.5192, 2.5278, 2.5258, 2.5323], device='cuda:1'), covar=tensor([0.1230, 0.0867, 0.2402, 0.2777, 0.1045, 0.1167, 0.1556, 0.0960], device='cuda:1'), in_proj_covar=tensor([0.0371, 0.0259, 0.0430, 0.0542, 0.0323, 0.0424, 0.0385, 0.0361], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 02:56:25,656 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90588.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:57:03,089 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0124, 2.0322, 2.0570, 2.1154, 2.0376, 1.6162, 1.3010, 1.8166], device='cuda:1'), covar=tensor([0.0577, 0.0465, 0.0619, 0.0347, 0.0499, 0.1494, 0.2328, 0.0448], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0171, 0.0144, 0.0141, 0.0203, 0.0137, 0.0159, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 02:57:12,299 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6380, 2.0581, 2.6464, 2.7348, 2.5720, 2.0357, 2.7283, 2.2035], device='cuda:1'), covar=tensor([0.0375, 0.0883, 0.0576, 0.0415, 0.0497, 0.1254, 0.0374, 0.0783], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0252, 0.0368, 0.0321, 0.0263, 0.0299, 0.0298, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 02:57:36,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.056e+02 2.705e+02 3.388e+02 5.902e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-08 02:57:50,343 INFO [train.py:873] (1/4) Epoch 12, batch 7500, loss[loss=0.1214, simple_loss=0.157, pruned_loss=0.04288, over 14192.00 frames. ], tot_loss[loss=0.1239, simple_loss=0.1546, pruned_loss=0.04657, over 1971844.90 frames. ], batch size: 69, lr: 6.46e-03, grad_scale: 8.0 2022-12-08 02:58:10,925 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:58:27,543 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 02:58:30,075 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2022-12-08 02:59:13,082 INFO [train.py:873] (1/4) Epoch 13, batch 0, loss[loss=0.1538, simple_loss=0.1862, pruned_loss=0.06069, over 14158.00 frames. ], tot_loss[loss=0.1538, simple_loss=0.1862, pruned_loss=0.06069, over 14158.00 frames. ], batch size: 84, lr: 6.21e-03, grad_scale: 8.0 2022-12-08 02:59:13,082 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 02:59:18,080 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3352, 2.4468, 2.3860, 2.6248, 2.0963, 2.5003, 2.1628, 1.5561], device='cuda:1'), covar=tensor([0.0656, 0.0687, 0.0707, 0.0471, 0.0932, 0.0483, 0.0911, 0.2035], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0080, 0.0064, 0.0067, 0.0093, 0.0078, 0.0094, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 02:59:20,362 INFO [train.py:905] (1/4) Epoch 13, validation: loss=0.1364, simple_loss=0.1777, pruned_loss=0.04756, over 857387.00 frames. 2022-12-08 02:59:20,363 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 02:59:41,454 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 6.584e+01 1.659e+02 2.772e+02 3.641e+02 1.065e+03, threshold=5.544e+02, percent-clipped=8.0 2022-12-08 02:59:57,623 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90786.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:09,216 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8430, 1.5762, 3.4880, 3.2486, 3.2749, 3.5159, 2.8313, 3.4609], device='cuda:1'), covar=tensor([0.1422, 0.1505, 0.0123, 0.0257, 0.0258, 0.0136, 0.0322, 0.0154], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0156, 0.0126, 0.0164, 0.0142, 0.0139, 0.0119, 0.0118], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:00:24,317 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90815.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:40,597 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90834.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:50,553 INFO [train.py:873] (1/4) Epoch 13, batch 100, loss[loss=0.1018, simple_loss=0.1388, pruned_loss=0.03237, over 14095.00 frames. ], tot_loss[loss=0.1222, simple_loss=0.1551, pruned_loss=0.04468, over 892339.44 frames. ], batch size: 22, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:00:54,353 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90849.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:56,704 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:01:06,886 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8260, 1.9989, 2.6258, 2.2495, 2.6690, 2.5702, 2.5249, 2.3180], device='cuda:1'), covar=tensor([0.0638, 0.2632, 0.0859, 0.1720, 0.0631, 0.1092, 0.0793, 0.1521], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0315, 0.0399, 0.0304, 0.0378, 0.0321, 0.0365, 0.0308], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:01:09,979 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.335e+02 2.702e+02 3.282e+02 9.825e+02, threshold=5.404e+02, percent-clipped=3.0 2022-12-08 03:01:27,459 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:01:34,181 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 03:01:37,929 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0972, 1.1420, 0.9986, 1.1012, 1.1479, 0.7333, 1.0180, 1.1160], device='cuda:1'), covar=tensor([0.0518, 0.0709, 0.0366, 0.0537, 0.0358, 0.0536, 0.0868, 0.0632], device='cuda:1'), in_proj_covar=tensor([0.0028, 0.0028, 0.0030, 0.0027, 0.0028, 0.0040, 0.0028, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:01:46,381 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90910.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:01:58,357 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4936, 3.0141, 2.8734, 1.8533, 2.9877, 3.1907, 3.5255, 2.5852], device='cuda:1'), covar=tensor([0.0711, 0.1506, 0.1200, 0.2050, 0.1068, 0.0664, 0.0743, 0.1605], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0177, 0.0136, 0.0124, 0.0134, 0.0145, 0.0123, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 03:02:06,648 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2747, 4.0577, 3.9390, 4.3147, 3.9054, 3.6457, 4.3324, 4.1662], device='cuda:1'), covar=tensor([0.0600, 0.0828, 0.0828, 0.0613, 0.0732, 0.0638, 0.0532, 0.0710], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0131, 0.0141, 0.0150, 0.0139, 0.0116, 0.0158, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:02:09,078 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:02:16,740 INFO [train.py:873] (1/4) Epoch 13, batch 200, loss[loss=0.1729, simple_loss=0.1591, pruned_loss=0.09334, over 1229.00 frames. ], tot_loss[loss=0.1226, simple_loss=0.1537, pruned_loss=0.04572, over 1245832.44 frames. ], batch size: 100, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:02:36,076 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.982e+01 2.220e+02 2.675e+02 3.390e+02 6.533e+02, threshold=5.349e+02, percent-clipped=6.0 2022-12-08 03:03:11,313 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:03:22,216 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3825, 2.0762, 2.4756, 2.5026, 2.2863, 1.9718, 2.5018, 2.1384], device='cuda:1'), covar=tensor([0.0301, 0.0697, 0.0330, 0.0298, 0.0469, 0.1102, 0.0355, 0.0541], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0255, 0.0373, 0.0326, 0.0267, 0.0302, 0.0302, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:03:30,673 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 03:03:44,490 INFO [train.py:873] (1/4) Epoch 13, batch 300, loss[loss=0.1307, simple_loss=0.1613, pruned_loss=0.05009, over 14227.00 frames. ], tot_loss[loss=0.1223, simple_loss=0.1536, pruned_loss=0.04546, over 1536070.08 frames. ], batch size: 60, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:03:53,666 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:04:04,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.005e+02 2.425e+02 3.069e+02 6.147e+02, threshold=4.849e+02, percent-clipped=1.0 2022-12-08 03:04:40,605 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-08 03:04:46,089 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:13,069 INFO [train.py:873] (1/4) Epoch 13, batch 400, loss[loss=0.1136, simple_loss=0.1514, pruned_loss=0.03795, over 14230.00 frames. ], tot_loss[loss=0.1225, simple_loss=0.1536, pruned_loss=0.04565, over 1664779.27 frames. ], batch size: 69, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:05:19,295 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91152.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:28,721 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91163.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:33,349 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.284e+02 2.800e+02 3.397e+02 6.782e+02, threshold=5.601e+02, percent-clipped=4.0 2022-12-08 03:05:39,064 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5100, 1.3503, 2.7676, 1.3421, 2.6491, 2.6695, 1.9636, 2.7337], device='cuda:1'), covar=tensor([0.0427, 0.2846, 0.0487, 0.2275, 0.0536, 0.0629, 0.1061, 0.0458], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0155, 0.0158, 0.0166, 0.0167, 0.0175, 0.0132, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 03:05:42,524 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8713, 2.6824, 2.7325, 2.9170, 2.8286, 2.8455, 2.9495, 2.4553], device='cuda:1'), covar=tensor([0.0634, 0.1121, 0.0586, 0.0571, 0.0736, 0.0468, 0.0726, 0.0684], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0264, 0.0188, 0.0184, 0.0179, 0.0150, 0.0271, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:06:01,542 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91200.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:06:06,396 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91205.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:06:41,226 INFO [train.py:873] (1/4) Epoch 13, batch 500, loss[loss=0.1138, simple_loss=0.1468, pruned_loss=0.04045, over 14283.00 frames. ], tot_loss[loss=0.122, simple_loss=0.1534, pruned_loss=0.04528, over 1805241.59 frames. ], batch size: 63, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:06:54,648 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 03:07:01,831 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.197e+02 2.790e+02 3.555e+02 6.703e+02, threshold=5.580e+02, percent-clipped=3.0 2022-12-08 03:07:16,564 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91285.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:07:28,526 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5810, 1.9855, 3.6295, 2.5993, 3.4551, 1.9539, 2.7931, 3.5598], device='cuda:1'), covar=tensor([0.0678, 0.4413, 0.0505, 0.5581, 0.0901, 0.3516, 0.1431, 0.0529], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0214, 0.0207, 0.0290, 0.0230, 0.0214, 0.0212, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:07:30,528 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 03:08:08,839 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 03:08:09,276 INFO [train.py:873] (1/4) Epoch 13, batch 600, loss[loss=0.101, simple_loss=0.1455, pruned_loss=0.0282, over 14471.00 frames. ], tot_loss[loss=0.1229, simple_loss=0.1537, pruned_loss=0.04599, over 1799404.56 frames. ], batch size: 24, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:08:10,223 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:08:29,164 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 2.257e+02 2.792e+02 3.414e+02 7.089e+02, threshold=5.584e+02, percent-clipped=8.0 2022-12-08 03:09:05,402 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8052, 3.5448, 3.4226, 3.8141, 3.5925, 3.7220, 3.8140, 3.1375], device='cuda:1'), covar=tensor([0.0474, 0.0980, 0.0494, 0.0456, 0.0797, 0.0380, 0.0529, 0.0650], device='cuda:1'), in_proj_covar=tensor([0.0167, 0.0261, 0.0185, 0.0182, 0.0176, 0.0148, 0.0267, 0.0163], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:09:36,011 INFO [train.py:873] (1/4) Epoch 13, batch 700, loss[loss=0.1249, simple_loss=0.1597, pruned_loss=0.04503, over 14420.00 frames. ], tot_loss[loss=0.1222, simple_loss=0.1532, pruned_loss=0.04562, over 1794861.07 frames. ], batch size: 53, lr: 6.18e-03, grad_scale: 8.0 2022-12-08 03:09:55,978 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.741e+01 2.087e+02 2.616e+02 3.337e+02 7.440e+02, threshold=5.233e+02, percent-clipped=3.0 2022-12-08 03:10:28,668 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:11:01,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91542.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:11:03,621 INFO [train.py:873] (1/4) Epoch 13, batch 800, loss[loss=0.139, simple_loss=0.1612, pruned_loss=0.05845, over 7799.00 frames. ], tot_loss[loss=0.1229, simple_loss=0.1537, pruned_loss=0.046, over 1849323.58 frames. ], batch size: 100, lr: 6.18e-03, grad_scale: 8.0 2022-12-08 03:11:08,947 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3530, 1.2618, 1.4085, 1.3714, 1.5841, 0.9234, 1.2937, 1.3263], device='cuda:1'), covar=tensor([0.0607, 0.0695, 0.0607, 0.0629, 0.0534, 0.0868, 0.0831, 0.0520], device='cuda:1'), in_proj_covar=tensor([0.0029, 0.0029, 0.0030, 0.0027, 0.0029, 0.0041, 0.0029, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:11:10,469 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91553.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:11:23,656 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.089e+02 2.194e+02 2.551e+02 3.061e+02 6.051e+02, threshold=5.101e+02, percent-clipped=1.0 2022-12-08 03:11:31,163 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8859, 0.8922, 0.8996, 0.8660, 0.8584, 0.5130, 0.5618, 0.7900], device='cuda:1'), covar=tensor([0.0096, 0.0102, 0.0088, 0.0103, 0.0106, 0.0233, 0.0143, 0.0181], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:11:40,677 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7143, 1.8495, 3.7218, 2.5404, 3.5375, 1.8671, 2.6924, 3.6054], device='cuda:1'), covar=tensor([0.0904, 0.4900, 0.0767, 0.6170, 0.1013, 0.4004, 0.1559, 0.0664], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0215, 0.0209, 0.0291, 0.0233, 0.0216, 0.0213, 0.0215], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0006, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:11:50,506 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7733, 0.8144, 0.6681, 0.8489, 0.8974, 0.2484, 0.7054, 0.8558], device='cuda:1'), covar=tensor([0.0277, 0.0569, 0.0384, 0.0334, 0.0344, 0.0295, 0.0814, 0.0529], device='cuda:1'), in_proj_covar=tensor([0.0029, 0.0029, 0.0030, 0.0028, 0.0029, 0.0041, 0.0029, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:11:54,251 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91603.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:12:24,200 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2937, 2.0936, 2.5737, 1.6161, 1.7889, 2.3924, 1.3284, 2.3313], device='cuda:1'), covar=tensor([0.0883, 0.1354, 0.0679, 0.2160, 0.2137, 0.0749, 0.3485, 0.0859], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0098, 0.0091, 0.0098, 0.0115, 0.0085, 0.0123, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:12:27,368 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:12:29,460 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3261, 3.1390, 2.3931, 3.3987, 3.2576, 3.2944, 3.0082, 2.4658], device='cuda:1'), covar=tensor([0.0884, 0.1406, 0.3493, 0.0652, 0.0898, 0.1250, 0.1263, 0.3170], device='cuda:1'), in_proj_covar=tensor([0.0269, 0.0297, 0.0270, 0.0262, 0.0315, 0.0297, 0.0258, 0.0249], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:12:30,940 INFO [train.py:873] (1/4) Epoch 13, batch 900, loss[loss=0.1093, simple_loss=0.1484, pruned_loss=0.03511, over 14281.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1546, pruned_loss=0.04645, over 1918008.85 frames. ], batch size: 80, lr: 6.18e-03, grad_scale: 16.0 2022-12-08 03:12:31,158 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9313, 2.0741, 2.7827, 2.2317, 2.8459, 2.7171, 2.6746, 2.4759], device='cuda:1'), covar=tensor([0.0727, 0.2511, 0.1035, 0.1629, 0.0629, 0.0966, 0.0754, 0.1574], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0319, 0.0402, 0.0306, 0.0379, 0.0324, 0.0364, 0.0308], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:12:52,063 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.371e+02 2.875e+02 3.801e+02 8.432e+02, threshold=5.750e+02, percent-clipped=7.0 2022-12-08 03:13:18,500 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1162, 1.9681, 2.0152, 2.1711, 2.0716, 1.9802, 2.1935, 1.8010], device='cuda:1'), covar=tensor([0.1227, 0.1415, 0.0779, 0.0808, 0.1007, 0.0765, 0.0920, 0.0777], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0266, 0.0189, 0.0185, 0.0180, 0.0149, 0.0271, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:13:44,867 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-08 03:13:58,929 INFO [train.py:873] (1/4) Epoch 13, batch 1000, loss[loss=0.1192, simple_loss=0.1532, pruned_loss=0.04257, over 14264.00 frames. ], tot_loss[loss=0.1231, simple_loss=0.1543, pruned_loss=0.04595, over 1975221.00 frames. ], batch size: 44, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:14:13,397 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4613, 3.7212, 3.4701, 3.5300, 2.5655, 3.5382, 3.4307, 1.8594], device='cuda:1'), covar=tensor([0.1532, 0.0532, 0.1378, 0.0705, 0.1115, 0.0679, 0.1054, 0.2195], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0079, 0.0064, 0.0066, 0.0093, 0.0078, 0.0093, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:14:19,310 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.021e+02 2.131e+02 2.597e+02 3.397e+02 6.255e+02, threshold=5.195e+02, percent-clipped=2.0 2022-12-08 03:15:26,315 INFO [train.py:873] (1/4) Epoch 13, batch 1100, loss[loss=0.1045, simple_loss=0.1407, pruned_loss=0.03408, over 13915.00 frames. ], tot_loss[loss=0.122, simple_loss=0.1534, pruned_loss=0.04526, over 2007510.46 frames. ], batch size: 20, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:15:47,466 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.763e+01 2.120e+02 2.631e+02 3.193e+02 5.598e+02, threshold=5.262e+02, percent-clipped=1.0 2022-12-08 03:16:02,214 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8370, 1.6342, 1.9406, 1.6818, 1.9389, 1.7915, 1.6514, 1.8384], device='cuda:1'), covar=tensor([0.0688, 0.1279, 0.0317, 0.0483, 0.0512, 0.0698, 0.0304, 0.0299], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0317, 0.0396, 0.0304, 0.0377, 0.0324, 0.0364, 0.0310], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:16:07,005 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91891.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:16:13,642 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91898.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:16:14,414 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2858, 3.0277, 3.0317, 3.2693, 3.1641, 3.2003, 3.3334, 2.7570], device='cuda:1'), covar=tensor([0.0603, 0.1176, 0.0532, 0.0553, 0.0843, 0.0473, 0.0639, 0.0681], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0265, 0.0189, 0.0186, 0.0181, 0.0149, 0.0273, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:16:36,904 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5512, 1.6997, 4.6393, 2.2154, 4.1744, 4.7578, 4.5245, 5.0785], device='cuda:1'), covar=tensor([0.0387, 0.4222, 0.0497, 0.2708, 0.0431, 0.0623, 0.0351, 0.0325], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0158, 0.0166, 0.0167, 0.0177, 0.0133, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:16:51,284 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91941.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:16:54,657 INFO [train.py:873] (1/4) Epoch 13, batch 1200, loss[loss=0.1994, simple_loss=0.1933, pruned_loss=0.1027, over 7794.00 frames. ], tot_loss[loss=0.1217, simple_loss=0.1533, pruned_loss=0.04502, over 1985213.17 frames. ], batch size: 100, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:17:00,819 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91952.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:03,182 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91955.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:15,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.334e+02 2.878e+02 3.521e+02 1.024e+03, threshold=5.756e+02, percent-clipped=7.0 2022-12-08 03:17:26,272 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91981.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:17:33,105 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:57,369 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92016.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:57,715 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 03:18:19,853 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92042.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:18:22,476 INFO [train.py:873] (1/4) Epoch 13, batch 1300, loss[loss=0.1454, simple_loss=0.1475, pruned_loss=0.07171, over 2659.00 frames. ], tot_loss[loss=0.1225, simple_loss=0.1535, pruned_loss=0.04571, over 1911263.52 frames. ], batch size: 100, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:18:44,122 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.023e+02 2.102e+02 2.497e+02 3.213e+02 6.107e+02, threshold=4.993e+02, percent-clipped=1.0 2022-12-08 03:19:15,910 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1244, 2.3460, 2.3391, 2.4102, 2.0734, 2.3865, 2.2865, 1.3337], device='cuda:1'), covar=tensor([0.1207, 0.0765, 0.0868, 0.0712, 0.1044, 0.0679, 0.1233, 0.2374], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0081, 0.0065, 0.0067, 0.0095, 0.0080, 0.0095, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:19:52,207 INFO [train.py:873] (1/4) Epoch 13, batch 1400, loss[loss=0.1175, simple_loss=0.1507, pruned_loss=0.04218, over 14227.00 frames. ], tot_loss[loss=0.123, simple_loss=0.1541, pruned_loss=0.04595, over 1933794.90 frames. ], batch size: 94, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:19:55,473 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-08 03:20:13,182 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.362e+02 2.970e+02 3.546e+02 6.469e+02, threshold=5.940e+02, percent-clipped=3.0 2022-12-08 03:20:32,316 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3563, 5.2152, 4.9243, 5.3682, 4.9001, 4.6182, 5.4153, 5.1485], device='cuda:1'), covar=tensor([0.0639, 0.0660, 0.0714, 0.0487, 0.0625, 0.0528, 0.0485, 0.0622], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0130, 0.0139, 0.0150, 0.0138, 0.0117, 0.0158, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:20:39,448 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92198.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:20:54,148 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9426, 1.4717, 3.8531, 1.7866, 3.8568, 4.0433, 2.9909, 4.3340], device='cuda:1'), covar=tensor([0.0220, 0.3090, 0.0446, 0.2107, 0.0409, 0.0382, 0.0659, 0.0154], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0156, 0.0159, 0.0167, 0.0168, 0.0177, 0.0134, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:21:21,063 INFO [train.py:873] (1/4) Epoch 13, batch 1500, loss[loss=0.1534, simple_loss=0.1541, pruned_loss=0.07629, over 2700.00 frames. ], tot_loss[loss=0.1224, simple_loss=0.1534, pruned_loss=0.04564, over 1938809.11 frames. ], batch size: 100, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:21:22,059 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92246.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:21:22,846 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:21:41,369 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.124e+02 2.104e+02 2.663e+02 3.435e+02 6.153e+02, threshold=5.325e+02, percent-clipped=2.0 2022-12-08 03:22:18,312 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92311.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:22:30,445 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5245, 1.9817, 3.5783, 2.5051, 3.4129, 1.9272, 2.7854, 3.4045], device='cuda:1'), covar=tensor([0.0718, 0.4557, 0.0512, 0.6184, 0.0920, 0.3489, 0.1356, 0.0523], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0214, 0.0210, 0.0289, 0.0231, 0.0215, 0.0214, 0.0214], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:22:40,877 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92337.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:22:48,257 INFO [train.py:873] (1/4) Epoch 13, batch 1600, loss[loss=0.1039, simple_loss=0.1409, pruned_loss=0.03346, over 14244.00 frames. ], tot_loss[loss=0.122, simple_loss=0.1531, pruned_loss=0.0454, over 1942343.75 frames. ], batch size: 25, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:22:58,914 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-08 03:23:08,502 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.214e+02 2.735e+02 3.327e+02 6.728e+02, threshold=5.470e+02, percent-clipped=2.0 2022-12-08 03:23:18,276 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 03:23:54,500 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8942, 4.0406, 4.2695, 3.5951, 4.1112, 4.2324, 1.9211, 3.8230], device='cuda:1'), covar=tensor([0.0303, 0.0342, 0.0329, 0.0635, 0.0303, 0.0269, 0.2836, 0.0291], device='cuda:1'), in_proj_covar=tensor([0.0159, 0.0167, 0.0141, 0.0139, 0.0199, 0.0133, 0.0154, 0.0185], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:24:14,890 INFO [train.py:873] (1/4) Epoch 13, batch 1700, loss[loss=0.1664, simple_loss=0.1509, pruned_loss=0.09096, over 1220.00 frames. ], tot_loss[loss=0.1215, simple_loss=0.153, pruned_loss=0.04499, over 1921846.21 frames. ], batch size: 100, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:24:31,065 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2161, 2.1429, 2.6098, 1.5160, 1.7462, 2.3867, 1.3441, 2.2601], device='cuda:1'), covar=tensor([0.1041, 0.1365, 0.0633, 0.2339, 0.2341, 0.0826, 0.3235, 0.1163], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0098, 0.0090, 0.0098, 0.0115, 0.0085, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:24:36,007 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.177e+02 2.526e+02 3.126e+02 5.467e+02, threshold=5.052e+02, percent-clipped=0.0 2022-12-08 03:25:42,443 INFO [train.py:873] (1/4) Epoch 13, batch 1800, loss[loss=0.09932, simple_loss=0.1405, pruned_loss=0.02909, over 14293.00 frames. ], tot_loss[loss=0.1221, simple_loss=0.1535, pruned_loss=0.04534, over 1961340.01 frames. ], batch size: 39, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:25:44,265 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92547.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:02,905 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.175e+02 2.080e+02 2.746e+02 3.507e+02 5.512e+02, threshold=5.493e+02, percent-clipped=3.0 2022-12-08 03:26:25,748 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92595.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:39,638 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92611.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:52,301 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:27:02,570 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92637.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:27:09,284 INFO [train.py:873] (1/4) Epoch 13, batch 1900, loss[loss=0.1351, simple_loss=0.1514, pruned_loss=0.05936, over 8608.00 frames. ], tot_loss[loss=0.1222, simple_loss=0.1538, pruned_loss=0.04531, over 2004734.88 frames. ], batch size: 100, lr: 6.14e-03, grad_scale: 4.0 2022-12-08 03:27:21,760 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:27:24,710 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6910, 2.3542, 3.0468, 1.9750, 2.0594, 2.7190, 1.4864, 2.7765], device='cuda:1'), covar=tensor([0.0976, 0.1168, 0.0600, 0.2049, 0.2120, 0.0800, 0.3524, 0.0834], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0096, 0.0090, 0.0097, 0.0113, 0.0084, 0.0121, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:27:31,531 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.202e+02 2.634e+02 3.218e+02 6.502e+02, threshold=5.268e+02, percent-clipped=3.0 2022-12-08 03:27:33,329 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0367, 3.8177, 3.7076, 4.0727, 3.9081, 3.5890, 4.1533, 3.4491], device='cuda:1'), covar=tensor([0.0530, 0.0877, 0.0443, 0.0447, 0.0733, 0.1300, 0.0503, 0.0561], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0269, 0.0191, 0.0188, 0.0183, 0.0150, 0.0275, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:27:44,640 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92685.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:27:45,532 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:28:15,704 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:28:18,467 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-08 03:28:37,930 INFO [train.py:873] (1/4) Epoch 13, batch 2000, loss[loss=0.1154, simple_loss=0.1495, pruned_loss=0.04059, over 14316.00 frames. ], tot_loss[loss=0.1219, simple_loss=0.1536, pruned_loss=0.04514, over 1994023.95 frames. ], batch size: 66, lr: 6.14e-03, grad_scale: 8.0 2022-12-08 03:28:59,569 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 03:28:59,687 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.366e+02 2.802e+02 3.411e+02 7.982e+02, threshold=5.603e+02, percent-clipped=8.0 2022-12-08 03:29:10,389 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:29:30,309 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3434, 1.2792, 1.2977, 1.3908, 1.5901, 0.9025, 1.2539, 1.2306], device='cuda:1'), covar=tensor([0.0703, 0.0807, 0.0625, 0.0571, 0.0403, 0.0988, 0.0721, 0.0765], device='cuda:1'), in_proj_covar=tensor([0.0029, 0.0029, 0.0031, 0.0027, 0.0029, 0.0041, 0.0029, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:29:33,006 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7845, 1.7406, 1.9453, 1.8458, 1.7778, 1.5975, 1.6047, 1.1471], device='cuda:1'), covar=tensor([0.0222, 0.0294, 0.0274, 0.0200, 0.0210, 0.0306, 0.0300, 0.0454], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:29:49,269 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0819, 1.0759, 1.2584, 0.9971, 0.8334, 0.8350, 0.9322, 0.8857], device='cuda:1'), covar=tensor([0.0208, 0.0228, 0.0159, 0.0218, 0.0267, 0.0487, 0.0274, 0.0425], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:29:54,456 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:29:58,535 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8410, 1.4385, 1.8328, 1.2601, 1.4767, 1.8703, 1.5456, 1.5652], device='cuda:1'), covar=tensor([0.0657, 0.0758, 0.0666, 0.0789, 0.1522, 0.0623, 0.0593, 0.1545], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0178, 0.0137, 0.0126, 0.0138, 0.0147, 0.0124, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 03:30:06,190 INFO [train.py:873] (1/4) Epoch 13, batch 2100, loss[loss=0.09774, simple_loss=0.1418, pruned_loss=0.02683, over 14391.00 frames. ], tot_loss[loss=0.1215, simple_loss=0.153, pruned_loss=0.04503, over 1988048.33 frames. ], batch size: 41, lr: 6.14e-03, grad_scale: 8.0 2022-12-08 03:30:07,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5029, 4.2122, 3.9299, 4.1053, 4.2969, 4.3688, 4.4528, 4.4380], device='cuda:1'), covar=tensor([0.0674, 0.0510, 0.1853, 0.2371, 0.0632, 0.0740, 0.0787, 0.0728], device='cuda:1'), in_proj_covar=tensor([0.0379, 0.0265, 0.0448, 0.0562, 0.0334, 0.0435, 0.0389, 0.0371], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:30:28,667 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.062e+02 2.147e+02 2.777e+02 3.486e+02 6.981e+02, threshold=5.554e+02, percent-clipped=1.0 2022-12-08 03:30:32,372 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7455, 5.2437, 5.1761, 5.6371, 5.2793, 4.6890, 5.6703, 4.6687], device='cuda:1'), covar=tensor([0.0325, 0.0953, 0.0301, 0.0431, 0.0715, 0.0433, 0.0485, 0.0489], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0267, 0.0191, 0.0188, 0.0182, 0.0151, 0.0275, 0.0167], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:30:48,193 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:30:54,307 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:21,971 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92931.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:34,898 INFO [train.py:873] (1/4) Epoch 13, batch 2200, loss[loss=0.1477, simple_loss=0.1758, pruned_loss=0.05981, over 10395.00 frames. ], tot_loss[loss=0.1219, simple_loss=0.1532, pruned_loss=0.0453, over 1985360.79 frames. ], batch size: 100, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:31:47,533 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92960.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:56,268 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.331e+02 2.968e+02 3.670e+02 7.983e+02, threshold=5.936e+02, percent-clipped=6.0 2022-12-08 03:32:01,006 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92975.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:06,310 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:12,610 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7550, 1.3327, 2.9980, 1.4964, 3.0141, 2.8950, 2.0078, 3.0593], device='cuda:1'), covar=tensor([0.0434, 0.3829, 0.0495, 0.2527, 0.0523, 0.0643, 0.1107, 0.0417], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0157, 0.0160, 0.0167, 0.0169, 0.0177, 0.0133, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:32:16,154 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:31,659 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93010.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:35,432 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.56 vs. limit=2.0 2022-12-08 03:32:55,179 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:55,197 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:02,672 INFO [train.py:873] (1/4) Epoch 13, batch 2300, loss[loss=0.141, simple_loss=0.158, pruned_loss=0.06204, over 3860.00 frames. ], tot_loss[loss=0.1218, simple_loss=0.153, pruned_loss=0.04529, over 1934051.88 frames. ], batch size: 100, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:33:25,583 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.107e+02 2.508e+02 3.191e+02 7.606e+02, threshold=5.016e+02, percent-clipped=1.0 2022-12-08 03:33:26,655 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93071.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:30,942 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93076.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:47,300 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2671, 3.4162, 3.5484, 3.2997, 3.4607, 3.1933, 1.4523, 3.2166], device='cuda:1'), covar=tensor([0.0368, 0.0357, 0.0358, 0.0411, 0.0337, 0.0622, 0.3093, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0162, 0.0169, 0.0142, 0.0139, 0.0200, 0.0135, 0.0157, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 03:33:49,194 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93097.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:34:05,035 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:34:31,777 INFO [train.py:873] (1/4) Epoch 13, batch 2400, loss[loss=0.1012, simple_loss=0.1421, pruned_loss=0.0302, over 13971.00 frames. ], tot_loss[loss=0.1226, simple_loss=0.1537, pruned_loss=0.04581, over 1936475.95 frames. ], batch size: 19, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:34:43,073 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3036, 3.1956, 3.9844, 2.7028, 2.5585, 3.4090, 1.9118, 3.3144], device='cuda:1'), covar=tensor([0.0999, 0.1025, 0.0547, 0.1947, 0.2057, 0.1008, 0.3503, 0.1573], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0096, 0.0089, 0.0097, 0.0113, 0.0085, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:34:50,258 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1998, 2.6316, 3.9074, 2.8817, 3.8938, 3.8265, 3.7429, 3.3467], device='cuda:1'), covar=tensor([0.0651, 0.2965, 0.1065, 0.2135, 0.0973, 0.0989, 0.1691, 0.2092], device='cuda:1'), in_proj_covar=tensor([0.0346, 0.0309, 0.0391, 0.0299, 0.0373, 0.0318, 0.0358, 0.0304], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:34:53,123 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.312e+02 2.703e+02 3.635e+02 6.826e+02, threshold=5.406e+02, percent-clipped=5.0 2022-12-08 03:34:58,926 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:35:09,390 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93187.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:35:39,953 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5963, 1.4897, 1.5757, 1.5050, 1.4436, 1.2580, 1.2139, 1.0377], device='cuda:1'), covar=tensor([0.0192, 0.0400, 0.0246, 0.0208, 0.0259, 0.0351, 0.0251, 0.0440], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:35:59,849 INFO [train.py:873] (1/4) Epoch 13, batch 2500, loss[loss=0.131, simple_loss=0.1324, pruned_loss=0.06483, over 2657.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1541, pruned_loss=0.0467, over 1851798.59 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:36:08,979 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93255.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:22,499 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.321e+02 2.910e+02 3.520e+02 9.929e+02, threshold=5.819e+02, percent-clipped=5.0 2022-12-08 03:36:32,087 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:37,131 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:38,215 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0292, 1.7655, 1.7762, 1.9522, 2.0550, 1.7423, 1.6998, 1.6076], device='cuda:1'), covar=tensor([0.0302, 0.0497, 0.0494, 0.0274, 0.0237, 0.0452, 0.0380, 0.0551], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0027, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:37:09,929 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8457, 0.7622, 0.7769, 0.7903, 0.7120, 0.4726, 0.6485, 0.6865], device='cuda:1'), covar=tensor([0.0150, 0.0163, 0.0138, 0.0152, 0.0180, 0.0347, 0.0210, 0.0287], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 03:37:15,080 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:16,827 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93331.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:18,389 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2022-12-08 03:37:28,915 INFO [train.py:873] (1/4) Epoch 13, batch 2600, loss[loss=0.1069, simple_loss=0.1447, pruned_loss=0.03453, over 12752.00 frames. ], tot_loss[loss=0.1216, simple_loss=0.1528, pruned_loss=0.0452, over 1857808.34 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:37:45,281 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.88 vs. limit=5.0 2022-12-08 03:37:47,475 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93366.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:50,652 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.180e+02 2.788e+02 3.347e+02 5.990e+02, threshold=5.575e+02, percent-clipped=1.0 2022-12-08 03:37:55,980 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:38:00,226 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9738, 2.3508, 3.9443, 4.1344, 3.9887, 2.4068, 4.0673, 3.2504], device='cuda:1'), covar=tensor([0.0389, 0.0981, 0.0991, 0.0379, 0.0399, 0.1457, 0.0416, 0.0770], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0256, 0.0371, 0.0327, 0.0266, 0.0303, 0.0305, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:38:01,925 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6141, 2.4125, 3.4757, 2.6768, 3.4526, 3.3329, 3.2600, 2.8548], device='cuda:1'), covar=tensor([0.1040, 0.2867, 0.0971, 0.1985, 0.1021, 0.1121, 0.1502, 0.1882], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0311, 0.0393, 0.0298, 0.0376, 0.0319, 0.0359, 0.0304], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:38:10,515 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93392.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:38:39,209 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93424.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:38:57,479 INFO [train.py:873] (1/4) Epoch 13, batch 2700, loss[loss=0.148, simple_loss=0.1746, pruned_loss=0.06069, over 9494.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1527, pruned_loss=0.04505, over 1902086.43 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:39:19,854 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.738e+01 2.118e+02 2.554e+02 3.202e+02 5.751e+02, threshold=5.109e+02, percent-clipped=2.0 2022-12-08 03:39:20,919 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:39:31,788 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 03:39:34,800 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93487.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:39:53,868 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6024, 3.4467, 4.2652, 3.0643, 2.6231, 3.5126, 2.0859, 3.6072], device='cuda:1'), covar=tensor([0.0894, 0.1038, 0.0691, 0.1292, 0.1940, 0.0859, 0.3326, 0.1180], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0097, 0.0090, 0.0098, 0.0115, 0.0085, 0.0123, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:39:54,768 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8117, 1.6692, 3.9964, 3.7448, 3.7705, 4.0487, 3.5713, 4.0809], device='cuda:1'), covar=tensor([0.1608, 0.1511, 0.0114, 0.0209, 0.0216, 0.0127, 0.0187, 0.0113], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0128, 0.0167, 0.0146, 0.0142, 0.0122, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:40:13,080 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 03:40:16,922 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93535.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:40:25,981 INFO [train.py:873] (1/4) Epoch 13, batch 2800, loss[loss=0.1251, simple_loss=0.1478, pruned_loss=0.05118, over 4951.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1525, pruned_loss=0.04456, over 1930116.04 frames. ], batch size: 100, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:40:30,239 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2022-12-08 03:40:35,189 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:40:47,337 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.714e+01 2.417e+02 3.010e+02 3.786e+02 1.160e+03, threshold=6.020e+02, percent-clipped=14.0 2022-12-08 03:40:49,183 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2124, 1.8822, 2.2454, 1.4973, 1.9461, 2.2179, 2.1886, 1.9403], device='cuda:1'), covar=tensor([0.0952, 0.0705, 0.0916, 0.1509, 0.1276, 0.0877, 0.0676, 0.1500], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0173, 0.0136, 0.0124, 0.0136, 0.0145, 0.0123, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 03:41:02,953 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93587.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:16,934 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:26,037 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9311, 5.8385, 5.5063, 6.0616, 5.5125, 5.2216, 6.0430, 5.8460], device='cuda:1'), covar=tensor([0.0575, 0.0660, 0.0659, 0.0356, 0.0695, 0.0375, 0.0536, 0.0487], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0130, 0.0139, 0.0150, 0.0139, 0.0116, 0.0157, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:41:41,614 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93631.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:44,877 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93635.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:49,954 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3351, 2.1312, 2.8194, 1.6813, 1.8112, 2.5236, 1.4815, 2.3772], device='cuda:1'), covar=tensor([0.1280, 0.1649, 0.0760, 0.2589, 0.2513, 0.0924, 0.3577, 0.1181], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0098, 0.0090, 0.0098, 0.0116, 0.0086, 0.0123, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:41:53,319 INFO [train.py:873] (1/4) Epoch 13, batch 2900, loss[loss=0.1752, simple_loss=0.1593, pruned_loss=0.09558, over 1284.00 frames. ], tot_loss[loss=0.1204, simple_loss=0.152, pruned_loss=0.04438, over 1918232.85 frames. ], batch size: 100, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:42:11,948 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93666.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:15,457 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.101e+02 2.613e+02 3.235e+02 8.582e+02, threshold=5.226e+02, percent-clipped=1.0 2022-12-08 03:42:23,189 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:28,098 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7801, 4.5619, 4.4650, 4.8650, 4.4621, 4.1672, 4.8922, 4.7189], device='cuda:1'), covar=tensor([0.0605, 0.0730, 0.0691, 0.0451, 0.0607, 0.0555, 0.0504, 0.0577], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0130, 0.0139, 0.0150, 0.0139, 0.0115, 0.0157, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:42:34,265 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:53,839 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93714.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:55,619 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6214, 2.4669, 3.2912, 2.2387, 2.0724, 2.8401, 1.5866, 2.7726], device='cuda:1'), covar=tensor([0.1437, 0.1541, 0.0725, 0.2689, 0.2518, 0.1283, 0.3778, 0.1423], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0097, 0.0090, 0.0098, 0.0115, 0.0085, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:43:02,525 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1612, 2.0194, 2.1017, 2.1996, 2.0894, 2.1098, 2.2501, 1.8928], device='cuda:1'), covar=tensor([0.0900, 0.1568, 0.0817, 0.0911, 0.1058, 0.0768, 0.0860, 0.0743], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0271, 0.0193, 0.0191, 0.0185, 0.0154, 0.0277, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:43:07,569 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7642, 3.0126, 3.0492, 3.1539, 2.3062, 3.1888, 2.8938, 1.4340], device='cuda:1'), covar=tensor([0.1369, 0.1026, 0.0743, 0.0477, 0.1004, 0.0425, 0.0899, 0.2420], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0082, 0.0066, 0.0068, 0.0097, 0.0082, 0.0096, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:43:16,176 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:43:21,028 INFO [train.py:873] (1/4) Epoch 13, batch 3000, loss[loss=0.1695, simple_loss=0.1468, pruned_loss=0.0961, over 1250.00 frames. ], tot_loss[loss=0.1202, simple_loss=0.1517, pruned_loss=0.0443, over 1915387.28 frames. ], batch size: 100, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:43:21,029 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 03:43:26,018 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5951, 2.7096, 2.7371, 2.9125, 2.4692, 2.8709, 2.4809, 1.7883], device='cuda:1'), covar=tensor([0.0791, 0.0889, 0.0659, 0.0417, 0.0726, 0.0514, 0.1010, 0.1954], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0083, 0.0066, 0.0068, 0.0097, 0.0082, 0.0096, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:43:29,438 INFO [train.py:905] (1/4) Epoch 13, validation: loss=0.1304, simple_loss=0.1697, pruned_loss=0.04555, over 857387.00 frames. 2022-12-08 03:43:29,438 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 03:43:52,064 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.240e+02 2.864e+02 3.755e+02 7.950e+02, threshold=5.728e+02, percent-clipped=3.0 2022-12-08 03:43:52,310 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:43:53,115 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:25,197 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6285, 1.4493, 3.6637, 1.5553, 3.5248, 3.7094, 2.6320, 3.9846], device='cuda:1'), covar=tensor([0.0237, 0.3168, 0.0382, 0.2334, 0.0637, 0.0462, 0.0916, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0158, 0.0160, 0.0171, 0.0171, 0.0179, 0.0136, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:44:32,293 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4367, 5.2099, 4.9611, 5.5089, 5.1639, 4.8056, 5.5635, 5.3437], device='cuda:1'), covar=tensor([0.0568, 0.0756, 0.0750, 0.0480, 0.0712, 0.0487, 0.0499, 0.0604], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0130, 0.0139, 0.0150, 0.0138, 0.0115, 0.0158, 0.0136], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:44:33,183 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9608, 1.7132, 1.9171, 1.9420, 1.8987, 1.9609, 1.9799, 1.7077], device='cuda:1'), covar=tensor([0.1709, 0.3080, 0.1828, 0.1682, 0.1994, 0.1342, 0.2148, 0.1813], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0273, 0.0194, 0.0192, 0.0186, 0.0154, 0.0279, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:44:35,622 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:43,782 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 03:44:46,155 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:47,107 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0434, 2.5401, 4.2028, 4.2283, 4.1182, 2.3421, 4.2529, 3.2400], device='cuda:1'), covar=tensor([0.0416, 0.1049, 0.0825, 0.0464, 0.0405, 0.1621, 0.0360, 0.0891], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0259, 0.0374, 0.0330, 0.0267, 0.0303, 0.0307, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:44:58,704 INFO [train.py:873] (1/4) Epoch 13, batch 3100, loss[loss=0.13, simple_loss=0.1593, pruned_loss=0.05032, over 14210.00 frames. ], tot_loss[loss=0.1213, simple_loss=0.1526, pruned_loss=0.04498, over 1903555.99 frames. ], batch size: 94, lr: 6.10e-03, grad_scale: 8.0 2022-12-08 03:45:19,448 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.395e+02 2.887e+02 3.517e+02 6.635e+02, threshold=5.774e+02, percent-clipped=1.0 2022-12-08 03:45:28,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2125, 2.3664, 2.4501, 2.5233, 2.1431, 2.5533, 2.3467, 1.4624], device='cuda:1'), covar=tensor([0.0900, 0.0939, 0.0755, 0.0483, 0.1016, 0.0550, 0.0925, 0.2198], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0081, 0.0066, 0.0068, 0.0096, 0.0081, 0.0096, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:45:30,255 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8512, 1.1118, 1.3223, 1.2603, 1.0995, 1.3136, 1.1123, 0.8613], device='cuda:1'), covar=tensor([0.1667, 0.1015, 0.0423, 0.0383, 0.1515, 0.0772, 0.1427, 0.1424], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0081, 0.0066, 0.0068, 0.0096, 0.0081, 0.0096, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:45:40,156 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4710, 1.1352, 2.0839, 1.8679, 1.8535, 2.0554, 1.4663, 2.0688], device='cuda:1'), covar=tensor([0.0873, 0.1380, 0.0204, 0.0469, 0.0614, 0.0267, 0.0601, 0.0275], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0160, 0.0127, 0.0165, 0.0144, 0.0141, 0.0121, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:45:50,896 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9230, 1.5503, 3.8051, 3.5017, 3.6082, 3.8561, 3.2784, 3.8297], device='cuda:1'), covar=tensor([0.1446, 0.1544, 0.0125, 0.0243, 0.0230, 0.0136, 0.0232, 0.0143], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0159, 0.0126, 0.0164, 0.0143, 0.0140, 0.0121, 0.0119], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:46:05,020 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 03:46:14,686 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2022-12-08 03:46:25,612 INFO [train.py:873] (1/4) Epoch 13, batch 3200, loss[loss=0.1253, simple_loss=0.1503, pruned_loss=0.05019, over 6969.00 frames. ], tot_loss[loss=0.121, simple_loss=0.1524, pruned_loss=0.04481, over 1905067.64 frames. ], batch size: 100, lr: 6.10e-03, grad_scale: 8.0 2022-12-08 03:46:33,872 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9355, 2.4017, 3.9424, 4.1242, 3.9455, 2.4635, 3.9687, 3.1084], device='cuda:1'), covar=tensor([0.0374, 0.0954, 0.0706, 0.0355, 0.0366, 0.1419, 0.0333, 0.0835], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0257, 0.0372, 0.0329, 0.0266, 0.0302, 0.0306, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:46:34,279 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2022-12-08 03:46:41,523 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-08 03:46:47,225 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0757, 2.1723, 1.9804, 2.1539, 1.8311, 2.0156, 2.1340, 2.0955], device='cuda:1'), covar=tensor([0.0891, 0.0979, 0.1222, 0.0841, 0.1421, 0.0890, 0.1012, 0.0835], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0131, 0.0140, 0.0150, 0.0139, 0.0116, 0.0158, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:46:48,614 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.002e+02 2.396e+02 3.046e+02 4.955e+02, threshold=4.792e+02, percent-clipped=0.0 2022-12-08 03:46:56,635 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4658, 4.2565, 3.8967, 4.0620, 4.2915, 4.3688, 4.4217, 4.4475], device='cuda:1'), covar=tensor([0.0883, 0.0486, 0.1884, 0.2629, 0.0802, 0.0806, 0.0998, 0.0794], device='cuda:1'), in_proj_covar=tensor([0.0378, 0.0266, 0.0444, 0.0564, 0.0333, 0.0438, 0.0388, 0.0372], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:46:57,613 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:47:43,202 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1354, 4.7247, 4.6433, 5.1444, 4.8054, 4.4719, 5.1331, 4.2526], device='cuda:1'), covar=tensor([0.0394, 0.0980, 0.0363, 0.0403, 0.0813, 0.0564, 0.0470, 0.0502], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0273, 0.0194, 0.0191, 0.0186, 0.0154, 0.0278, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:47:50,344 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94042.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:47:53,369 INFO [train.py:873] (1/4) Epoch 13, batch 3300, loss[loss=0.121, simple_loss=0.1535, pruned_loss=0.04424, over 14214.00 frames. ], tot_loss[loss=0.1212, simple_loss=0.1526, pruned_loss=0.04492, over 1910785.61 frames. ], batch size: 94, lr: 6.10e-03, grad_scale: 4.0 2022-12-08 03:47:54,408 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7623, 1.9844, 2.6757, 2.1781, 2.6537, 2.5999, 2.3246, 2.3213], device='cuda:1'), covar=tensor([0.0691, 0.2566, 0.0834, 0.1547, 0.0559, 0.1021, 0.0683, 0.1229], device='cuda:1'), in_proj_covar=tensor([0.0347, 0.0312, 0.0394, 0.0299, 0.0375, 0.0317, 0.0357, 0.0305], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:48:15,567 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.207e+02 2.819e+02 3.660e+02 9.376e+02, threshold=5.637e+02, percent-clipped=6.0 2022-12-08 03:48:17,344 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6079, 2.0191, 2.6071, 2.6639, 2.4673, 2.0281, 2.6914, 2.2142], device='cuda:1'), covar=tensor([0.0456, 0.1028, 0.0565, 0.0417, 0.0597, 0.1274, 0.0430, 0.0840], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0256, 0.0372, 0.0328, 0.0267, 0.0301, 0.0306, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:48:20,632 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7172, 5.5057, 5.1388, 5.7363, 5.2988, 5.1468, 5.8338, 5.5905], device='cuda:1'), covar=tensor([0.0704, 0.0714, 0.0896, 0.0683, 0.0886, 0.0486, 0.0568, 0.0776], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0131, 0.0140, 0.0151, 0.0139, 0.0117, 0.0159, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:49:03,579 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94126.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:49:19,899 INFO [train.py:873] (1/4) Epoch 13, batch 3400, loss[loss=0.1163, simple_loss=0.1412, pruned_loss=0.04567, over 14576.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1526, pruned_loss=0.04512, over 1934113.11 frames. ], batch size: 21, lr: 6.09e-03, grad_scale: 4.0 2022-12-08 03:49:42,649 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.362e+02 2.991e+02 3.918e+02 1.858e+03, threshold=5.982e+02, percent-clipped=6.0 2022-12-08 03:49:58,226 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7896, 5.6006, 5.1959, 5.8301, 5.3774, 5.2110, 5.9009, 5.6904], device='cuda:1'), covar=tensor([0.0503, 0.0506, 0.0783, 0.0353, 0.0585, 0.0288, 0.0408, 0.0433], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0130, 0.0140, 0.0150, 0.0138, 0.0116, 0.0158, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 03:50:19,747 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2695, 1.7341, 2.2565, 1.8932, 2.3585, 2.0415, 1.9815, 2.0560], device='cuda:1'), covar=tensor([0.0565, 0.1752, 0.0481, 0.0822, 0.0518, 0.0959, 0.0379, 0.0566], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0316, 0.0397, 0.0302, 0.0378, 0.0321, 0.0364, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:50:28,488 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94223.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:50:36,987 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7703, 5.2469, 5.2408, 5.7632, 5.3945, 4.6622, 5.7161, 4.6277], device='cuda:1'), covar=tensor([0.0349, 0.1228, 0.0367, 0.0417, 0.0769, 0.0421, 0.0434, 0.0590], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0273, 0.0193, 0.0189, 0.0184, 0.0153, 0.0276, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:50:47,830 INFO [train.py:873] (1/4) Epoch 13, batch 3500, loss[loss=0.1081, simple_loss=0.1468, pruned_loss=0.03465, over 14221.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.1522, pruned_loss=0.04435, over 1953416.46 frames. ], batch size: 60, lr: 6.09e-03, grad_scale: 4.0 2022-12-08 03:51:05,151 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8105, 2.2847, 3.3041, 2.1623, 2.0141, 2.8058, 1.4237, 2.6526], device='cuda:1'), covar=tensor([0.1027, 0.1527, 0.0617, 0.2296, 0.2243, 0.1149, 0.4033, 0.1203], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0096, 0.0089, 0.0096, 0.0113, 0.0084, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:51:07,753 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4339, 2.2055, 2.7258, 1.7620, 1.8297, 2.4725, 1.3906, 2.2995], device='cuda:1'), covar=tensor([0.0926, 0.1405, 0.0557, 0.1962, 0.2428, 0.0844, 0.3980, 0.0947], device='cuda:1'), in_proj_covar=tensor([0.0082, 0.0096, 0.0089, 0.0096, 0.0113, 0.0084, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 03:51:10,108 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.271e+02 2.788e+02 3.634e+02 1.227e+03, threshold=5.575e+02, percent-clipped=1.0 2022-12-08 03:51:16,984 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1374, 2.0114, 1.7964, 1.8659, 2.0813, 2.0772, 2.0757, 2.0649], device='cuda:1'), covar=tensor([0.1178, 0.0914, 0.2768, 0.2765, 0.1146, 0.1265, 0.1662, 0.1243], device='cuda:1'), in_proj_covar=tensor([0.0372, 0.0262, 0.0438, 0.0555, 0.0324, 0.0430, 0.0381, 0.0368], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:51:21,992 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94284.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:52:03,804 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7080, 3.3924, 3.2454, 2.4287, 3.1743, 3.4383, 3.8317, 3.0262], device='cuda:1'), covar=tensor([0.0542, 0.1295, 0.0831, 0.1435, 0.0782, 0.0656, 0.0697, 0.1024], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0177, 0.0139, 0.0126, 0.0137, 0.0147, 0.0124, 0.0138], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 03:52:08,032 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:52:14,716 INFO [train.py:873] (1/4) Epoch 13, batch 3600, loss[loss=0.1207, simple_loss=0.1582, pruned_loss=0.04161, over 14095.00 frames. ], tot_loss[loss=0.1219, simple_loss=0.1533, pruned_loss=0.04529, over 1966664.29 frames. ], batch size: 29, lr: 6.09e-03, grad_scale: 8.0 2022-12-08 03:52:17,390 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94348.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:52:37,864 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.011e+02 2.604e+02 3.365e+02 5.868e+02, threshold=5.208e+02, percent-clipped=1.0 2022-12-08 03:53:11,003 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94409.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:53:25,340 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94426.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:53:32,981 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7899, 1.1081, 1.2670, 1.1574, 1.1703, 1.2547, 1.0800, 0.8240], device='cuda:1'), covar=tensor([0.1873, 0.1021, 0.0385, 0.0555, 0.1414, 0.0814, 0.1484, 0.1554], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0083, 0.0066, 0.0070, 0.0097, 0.0083, 0.0097, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 03:53:34,490 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 03:53:41,356 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4982, 2.3334, 3.2517, 2.4808, 3.3275, 3.2456, 3.1650, 2.6895], device='cuda:1'), covar=tensor([0.0869, 0.2732, 0.1030, 0.2184, 0.0939, 0.0991, 0.1394, 0.1893], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0314, 0.0396, 0.0302, 0.0378, 0.0322, 0.0363, 0.0305], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:53:41,902 INFO [train.py:873] (1/4) Epoch 13, batch 3700, loss[loss=0.1371, simple_loss=0.1658, pruned_loss=0.05415, over 12739.00 frames. ], tot_loss[loss=0.1215, simple_loss=0.1529, pruned_loss=0.04503, over 1972828.67 frames. ], batch size: 100, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:54:03,058 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.07 vs. limit=5.0 2022-12-08 03:54:04,972 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.353e+02 2.765e+02 3.627e+02 6.924e+02, threshold=5.531e+02, percent-clipped=2.0 2022-12-08 03:54:06,723 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:54:36,789 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7399, 2.3918, 2.5499, 1.6546, 2.2680, 2.5181, 2.7358, 2.2025], device='cuda:1'), covar=tensor([0.0737, 0.0884, 0.1033, 0.1607, 0.1080, 0.0771, 0.0723, 0.1487], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0177, 0.0139, 0.0126, 0.0138, 0.0148, 0.0125, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 03:54:42,894 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7844, 2.5530, 4.8056, 3.2749, 4.5418, 2.3055, 3.4770, 4.5475], device='cuda:1'), covar=tensor([0.0428, 0.3677, 0.0285, 0.6517, 0.0512, 0.3355, 0.1222, 0.0403], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0209, 0.0207, 0.0283, 0.0224, 0.0209, 0.0210, 0.0210], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:55:09,007 INFO [train.py:873] (1/4) Epoch 13, batch 3800, loss[loss=0.1259, simple_loss=0.1552, pruned_loss=0.04831, over 6920.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1528, pruned_loss=0.04498, over 1922625.02 frames. ], batch size: 100, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:55:32,195 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.117e+02 2.466e+02 3.115e+02 4.944e+02, threshold=4.932e+02, percent-clipped=0.0 2022-12-08 03:55:38,483 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94579.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:56:29,395 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94637.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:56:36,274 INFO [train.py:873] (1/4) Epoch 13, batch 3900, loss[loss=0.1153, simple_loss=0.1486, pruned_loss=0.04098, over 14249.00 frames. ], tot_loss[loss=0.1202, simple_loss=0.1518, pruned_loss=0.04431, over 1903886.92 frames. ], batch size: 39, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:56:39,162 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8149, 3.5638, 3.4070, 3.8126, 3.5583, 3.7863, 3.8331, 3.2216], device='cuda:1'), covar=tensor([0.0436, 0.0931, 0.0517, 0.0458, 0.0874, 0.0318, 0.0571, 0.0574], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0272, 0.0192, 0.0187, 0.0183, 0.0151, 0.0274, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 03:56:59,461 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 2.050e+02 2.441e+02 3.382e+02 7.650e+02, threshold=4.881e+02, percent-clipped=6.0 2022-12-08 03:57:00,533 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7249, 1.4226, 3.7337, 3.5179, 3.5637, 3.8319, 3.0772, 3.7437], device='cuda:1'), covar=tensor([0.1780, 0.1925, 0.0195, 0.0341, 0.0322, 0.0206, 0.0375, 0.0211], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0160, 0.0127, 0.0167, 0.0145, 0.0140, 0.0122, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 03:57:11,476 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:57:18,259 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9346, 3.0059, 4.7049, 3.5060, 4.6599, 4.7955, 4.5999, 4.1013], device='cuda:1'), covar=tensor([0.0733, 0.2909, 0.0782, 0.1844, 0.0845, 0.0655, 0.1381, 0.1558], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0316, 0.0397, 0.0302, 0.0380, 0.0324, 0.0364, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 03:57:27,733 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94704.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:57:57,297 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2022-12-08 03:58:02,597 INFO [train.py:873] (1/4) Epoch 13, batch 4000, loss[loss=0.123, simple_loss=0.1424, pruned_loss=0.05183, over 6968.00 frames. ], tot_loss[loss=0.1193, simple_loss=0.1513, pruned_loss=0.04367, over 1931661.10 frames. ], batch size: 100, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 03:58:13,679 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 03:58:26,193 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.218e+02 2.831e+02 3.684e+02 6.338e+02, threshold=5.663e+02, percent-clipped=4.0 2022-12-08 03:59:29,380 INFO [train.py:873] (1/4) Epoch 13, batch 4100, loss[loss=0.1447, simple_loss=0.1565, pruned_loss=0.06641, over 5023.00 frames. ], tot_loss[loss=0.1194, simple_loss=0.1518, pruned_loss=0.04351, over 1999634.21 frames. ], batch size: 100, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 03:59:52,465 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 2.091e+02 2.510e+02 3.105e+02 7.353e+02, threshold=5.020e+02, percent-clipped=2.0 2022-12-08 03:59:58,799 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94879.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:00:00,641 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3974, 2.2028, 3.3769, 3.5807, 3.4645, 2.2161, 3.4636, 2.6384], device='cuda:1'), covar=tensor([0.0454, 0.0963, 0.0781, 0.0483, 0.0425, 0.1511, 0.0476, 0.1044], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0251, 0.0364, 0.0320, 0.0263, 0.0298, 0.0300, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:00:09,511 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8652, 0.6942, 0.8038, 0.9054, 0.7750, 0.3891, 0.6272, 0.8437], device='cuda:1'), covar=tensor([0.0148, 0.0133, 0.0133, 0.0110, 0.0164, 0.0329, 0.0209, 0.0243], device='cuda:1'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0027, 0.0022, 0.0027], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:00:11,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8050, 4.4295, 4.2900, 4.8015, 4.4777, 4.3266, 4.7912, 4.0749], device='cuda:1'), covar=tensor([0.0361, 0.1068, 0.0389, 0.0420, 0.0813, 0.0620, 0.0492, 0.0479], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0273, 0.0192, 0.0189, 0.0184, 0.0153, 0.0277, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:00:29,388 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2022-12-08 04:00:40,986 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94927.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:00:52,663 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2652, 2.1429, 2.4822, 1.6020, 1.7184, 2.1722, 1.3962, 2.1887], device='cuda:1'), covar=tensor([0.1057, 0.1433, 0.0797, 0.2551, 0.2804, 0.1201, 0.3550, 0.1150], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0097, 0.0090, 0.0097, 0.0114, 0.0085, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 04:00:56,134 INFO [train.py:873] (1/4) Epoch 13, batch 4200, loss[loss=0.1186, simple_loss=0.1575, pruned_loss=0.03989, over 14227.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.1525, pruned_loss=0.04399, over 1965645.60 frames. ], batch size: 57, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 04:01:10,952 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1607, 2.0040, 2.0993, 2.1637, 2.1398, 2.1038, 2.2227, 1.8773], device='cuda:1'), covar=tensor([0.0848, 0.1220, 0.0761, 0.0820, 0.0949, 0.0676, 0.0795, 0.0780], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0269, 0.0190, 0.0187, 0.0182, 0.0151, 0.0274, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:01:19,978 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.402e+02 2.842e+02 3.518e+02 1.114e+03, threshold=5.683e+02, percent-clipped=6.0 2022-12-08 04:01:48,003 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0753, 2.8132, 2.2543, 3.1894, 2.9648, 3.0694, 2.7125, 2.2451], device='cuda:1'), covar=tensor([0.1058, 0.1573, 0.3610, 0.0750, 0.1052, 0.1307, 0.1559, 0.3816], device='cuda:1'), in_proj_covar=tensor([0.0273, 0.0290, 0.0264, 0.0265, 0.0312, 0.0297, 0.0255, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:01:48,429 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-08 04:01:51,312 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95004.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:02:03,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9034, 3.6751, 3.3341, 2.6433, 3.3652, 3.6177, 3.9301, 3.2132], device='cuda:1'), covar=tensor([0.0442, 0.1025, 0.0763, 0.1187, 0.0697, 0.0566, 0.0762, 0.0930], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0177, 0.0139, 0.0126, 0.0138, 0.0150, 0.0125, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 04:02:26,455 INFO [train.py:873] (1/4) Epoch 13, batch 4300, loss[loss=0.122, simple_loss=0.131, pruned_loss=0.0565, over 2635.00 frames. ], tot_loss[loss=0.1207, simple_loss=0.1529, pruned_loss=0.04424, over 1961692.62 frames. ], batch size: 100, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:02:32,650 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95052.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:02:49,664 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.351e+02 2.797e+02 3.272e+02 6.145e+02, threshold=5.594e+02, percent-clipped=1.0 2022-12-08 04:03:18,348 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95105.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:03:53,023 INFO [train.py:873] (1/4) Epoch 13, batch 4400, loss[loss=0.1192, simple_loss=0.1574, pruned_loss=0.04054, over 14281.00 frames. ], tot_loss[loss=0.1212, simple_loss=0.1531, pruned_loss=0.04464, over 1965785.69 frames. ], batch size: 44, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:04:10,975 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95166.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:04:16,201 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.201e+02 2.204e+02 2.628e+02 3.192e+02 7.807e+02, threshold=5.256e+02, percent-clipped=2.0 2022-12-08 04:04:51,896 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2022-12-08 04:05:02,949 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9973, 3.8184, 3.4070, 2.7944, 3.4153, 3.7429, 4.1803, 3.2946], device='cuda:1'), covar=tensor([0.0449, 0.0892, 0.0883, 0.1152, 0.0874, 0.0547, 0.0446, 0.1000], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0179, 0.0141, 0.0128, 0.0139, 0.0150, 0.0125, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 04:05:20,305 INFO [train.py:873] (1/4) Epoch 13, batch 4500, loss[loss=0.1266, simple_loss=0.1577, pruned_loss=0.04779, over 14269.00 frames. ], tot_loss[loss=0.1195, simple_loss=0.1522, pruned_loss=0.0434, over 2036065.21 frames. ], batch size: 46, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:05:24,096 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9787, 2.1423, 1.9247, 2.1255, 1.8275, 2.0369, 2.0581, 2.0199], device='cuda:1'), covar=tensor([0.1076, 0.1065, 0.1444, 0.0923, 0.1399, 0.0926, 0.1218, 0.1005], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0133, 0.0143, 0.0152, 0.0140, 0.0117, 0.0162, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:05:33,383 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:05:40,389 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:05:43,635 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.015e+02 2.327e+02 3.103e+02 3.682e+02 7.295e+02, threshold=6.206e+02, percent-clipped=3.0 2022-12-08 04:06:26,020 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:33,399 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:41,665 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95339.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:42,536 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9480, 1.5747, 1.9782, 1.6620, 1.9998, 1.8723, 1.7137, 1.9402], device='cuda:1'), covar=tensor([0.0469, 0.1603, 0.0315, 0.0461, 0.0398, 0.0650, 0.0314, 0.0424], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0309, 0.0391, 0.0300, 0.0372, 0.0319, 0.0362, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:06:46,993 INFO [train.py:873] (1/4) Epoch 13, batch 4600, loss[loss=0.1073, simple_loss=0.1237, pruned_loss=0.04542, over 2611.00 frames. ], tot_loss[loss=0.1189, simple_loss=0.1515, pruned_loss=0.04321, over 1946088.97 frames. ], batch size: 100, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:07:10,470 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.077e+02 2.550e+02 3.185e+02 6.474e+02, threshold=5.099e+02, percent-clipped=1.0 2022-12-08 04:07:18,308 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0740, 3.1368, 3.2330, 3.0760, 3.1290, 3.0238, 1.4981, 2.9739], device='cuda:1'), covar=tensor([0.0369, 0.0375, 0.0382, 0.0420, 0.0352, 0.0779, 0.2711, 0.0356], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0173, 0.0144, 0.0142, 0.0202, 0.0137, 0.0160, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 04:07:35,117 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95400.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:08:14,530 INFO [train.py:873] (1/4) Epoch 13, batch 4700, loss[loss=0.1398, simple_loss=0.1686, pruned_loss=0.05543, over 10382.00 frames. ], tot_loss[loss=0.1198, simple_loss=0.1516, pruned_loss=0.04398, over 1903894.64 frames. ], batch size: 100, lr: 6.05e-03, grad_scale: 4.0 2022-12-08 04:08:28,736 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95461.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:08:39,145 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.071e+02 2.759e+02 3.394e+02 8.943e+02, threshold=5.519e+02, percent-clipped=4.0 2022-12-08 04:09:15,281 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7186, 2.7277, 2.0062, 2.8534, 2.6731, 2.6883, 2.4050, 2.1902], device='cuda:1'), covar=tensor([0.1176, 0.1225, 0.3336, 0.0838, 0.1095, 0.0878, 0.1625, 0.2799], device='cuda:1'), in_proj_covar=tensor([0.0272, 0.0292, 0.0267, 0.0264, 0.0313, 0.0297, 0.0256, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:09:41,414 INFO [train.py:873] (1/4) Epoch 13, batch 4800, loss[loss=0.1023, simple_loss=0.1481, pruned_loss=0.0283, over 14538.00 frames. ], tot_loss[loss=0.1215, simple_loss=0.1523, pruned_loss=0.04532, over 1857551.78 frames. ], batch size: 34, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:09:48,731 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1323, 2.0368, 1.8027, 1.8416, 2.0416, 2.0551, 2.0921, 2.0831], device='cuda:1'), covar=tensor([0.1109, 0.0839, 0.2631, 0.2471, 0.1228, 0.1148, 0.1458, 0.1164], device='cuda:1'), in_proj_covar=tensor([0.0375, 0.0267, 0.0443, 0.0553, 0.0331, 0.0431, 0.0381, 0.0371], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:10:05,764 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.302e+02 2.972e+02 3.662e+02 7.055e+02, threshold=5.944e+02, percent-clipped=3.0 2022-12-08 04:10:43,507 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:10:44,414 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:10:50,236 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:11:02,859 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3028, 2.3774, 1.9258, 2.4817, 2.3224, 2.3158, 2.1342, 2.0399], device='cuda:1'), covar=tensor([0.1007, 0.0891, 0.2249, 0.0688, 0.1190, 0.0731, 0.1600, 0.1267], device='cuda:1'), in_proj_covar=tensor([0.0273, 0.0295, 0.0269, 0.0267, 0.0314, 0.0300, 0.0258, 0.0247], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:11:08,087 INFO [train.py:873] (1/4) Epoch 13, batch 4900, loss[loss=0.1262, simple_loss=0.1629, pruned_loss=0.04473, over 14289.00 frames. ], tot_loss[loss=0.121, simple_loss=0.1524, pruned_loss=0.04484, over 1874594.35 frames. ], batch size: 25, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:11:18,405 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-08 04:11:25,267 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6951, 1.5344, 3.9111, 3.6392, 3.6466, 3.9734, 3.3112, 3.9182], device='cuda:1'), covar=tensor([0.1648, 0.1626, 0.0114, 0.0236, 0.0233, 0.0125, 0.0252, 0.0129], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0128, 0.0168, 0.0145, 0.0140, 0.0123, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 04:11:32,372 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.158e+02 2.610e+02 3.322e+02 1.003e+03, threshold=5.220e+02, percent-clipped=4.0 2022-12-08 04:11:34,137 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7525, 1.6952, 1.7927, 1.7743, 1.8422, 1.4292, 1.4838, 1.1045], device='cuda:1'), covar=tensor([0.0221, 0.0514, 0.0343, 0.0241, 0.0272, 0.0351, 0.0295, 0.0500], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0016, 0.0017, 0.0017, 0.0029, 0.0023, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:11:36,559 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:11:49,874 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7697, 4.1671, 4.4449, 4.7200, 4.3667, 4.1238, 4.8813, 4.0152], device='cuda:1'), covar=tensor([0.0867, 0.1960, 0.0780, 0.0939, 0.1353, 0.0921, 0.0850, 0.0854], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0272, 0.0191, 0.0188, 0.0186, 0.0152, 0.0277, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:11:51,493 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95695.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:12:15,732 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9094, 1.6670, 4.1991, 3.9422, 3.8660, 4.2701, 3.6424, 4.2753], device='cuda:1'), covar=tensor([0.1499, 0.1552, 0.0114, 0.0231, 0.0244, 0.0134, 0.0271, 0.0125], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0127, 0.0167, 0.0145, 0.0140, 0.0122, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 04:12:16,189 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-08 04:12:27,161 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 04:12:35,018 INFO [train.py:873] (1/4) Epoch 13, batch 5000, loss[loss=0.1283, simple_loss=0.1582, pruned_loss=0.04915, over 9502.00 frames. ], tot_loss[loss=0.1216, simple_loss=0.153, pruned_loss=0.04512, over 1932857.13 frames. ], batch size: 100, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:12:49,283 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95761.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:12:51,207 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2022-12-08 04:12:59,674 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.284e+02 2.864e+02 3.553e+02 7.569e+02, threshold=5.729e+02, percent-clipped=2.0 2022-12-08 04:12:59,835 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9712, 1.4975, 1.9589, 1.3219, 1.6538, 1.9990, 1.8468, 1.6896], device='cuda:1'), covar=tensor([0.0928, 0.0841, 0.0872, 0.1660, 0.1528, 0.0952, 0.0840, 0.1755], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0175, 0.0137, 0.0125, 0.0137, 0.0148, 0.0123, 0.0137], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 04:13:05,406 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5352, 5.3498, 4.9911, 5.5681, 5.2048, 5.0294, 5.5939, 5.4584], device='cuda:1'), covar=tensor([0.0659, 0.0726, 0.1010, 0.0562, 0.0695, 0.0503, 0.0568, 0.0661], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0133, 0.0143, 0.0153, 0.0140, 0.0117, 0.0161, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:13:17,297 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 04:13:19,959 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9561, 1.8988, 1.4377, 1.4773, 1.8751, 1.9061, 1.9402, 1.9290], device='cuda:1'), covar=tensor([0.1753, 0.1200, 0.4306, 0.4168, 0.2028, 0.1823, 0.2110, 0.1732], device='cuda:1'), in_proj_covar=tensor([0.0377, 0.0269, 0.0444, 0.0557, 0.0330, 0.0431, 0.0383, 0.0368], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:13:31,718 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95809.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:13:34,747 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95812.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:13:46,599 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1423, 2.8858, 3.7554, 2.5507, 2.4599, 3.3063, 1.6818, 3.2062], device='cuda:1'), covar=tensor([0.0991, 0.1130, 0.0566, 0.1637, 0.1764, 0.0728, 0.3504, 0.1069], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0098, 0.0091, 0.0097, 0.0113, 0.0086, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 04:13:50,477 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0850, 2.4984, 5.0954, 3.5342, 4.8602, 2.2694, 3.8706, 4.8237], device='cuda:1'), covar=tensor([0.0350, 0.3629, 0.0234, 0.5722, 0.0468, 0.3224, 0.1104, 0.0297], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0208, 0.0208, 0.0284, 0.0226, 0.0211, 0.0208, 0.0210], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:13:56,872 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 04:14:03,053 INFO [train.py:873] (1/4) Epoch 13, batch 5100, loss[loss=0.13, simple_loss=0.1575, pruned_loss=0.05128, over 5964.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.1523, pruned_loss=0.04416, over 1968140.67 frames. ], batch size: 100, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:14:04,996 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95847.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:14:08,495 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1352, 2.1388, 2.3847, 1.4637, 1.6990, 2.1655, 1.3637, 2.1443], device='cuda:1'), covar=tensor([0.1248, 0.1468, 0.0775, 0.2664, 0.2635, 0.1071, 0.3366, 0.0978], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0098, 0.0091, 0.0098, 0.0114, 0.0087, 0.0122, 0.0089], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 04:14:27,856 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.050e+02 2.508e+02 3.173e+02 5.976e+02, threshold=5.017e+02, percent-clipped=1.0 2022-12-08 04:14:28,047 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95873.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:14:34,307 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 04:14:59,165 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95908.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:15:06,317 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:13,280 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:20,615 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4689, 1.0154, 2.0695, 1.8717, 1.8984, 2.0993, 1.5763, 2.0906], device='cuda:1'), covar=tensor([0.0697, 0.1206, 0.0215, 0.0425, 0.0527, 0.0209, 0.0484, 0.0271], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0166, 0.0144, 0.0138, 0.0121, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 04:15:31,806 INFO [train.py:873] (1/4) Epoch 13, batch 5200, loss[loss=0.1009, simple_loss=0.1405, pruned_loss=0.03063, over 14411.00 frames. ], tot_loss[loss=0.1206, simple_loss=0.1523, pruned_loss=0.04443, over 1944042.10 frames. ], batch size: 53, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:15:48,362 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95964.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:55,698 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:56,530 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 2.184e+02 2.660e+02 3.283e+02 4.676e+02, threshold=5.321e+02, percent-clipped=0.0 2022-12-08 04:15:56,654 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:02,973 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:15,832 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95995.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:39,806 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7689, 3.4799, 3.4937, 3.7841, 3.6016, 3.7450, 3.8427, 3.1503], device='cuda:1'), covar=tensor([0.0502, 0.1090, 0.0451, 0.0542, 0.0784, 0.0361, 0.0532, 0.0633], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0272, 0.0191, 0.0188, 0.0185, 0.0152, 0.0279, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:16:39,857 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8377, 1.4462, 3.2029, 2.9733, 3.0980, 3.2081, 2.4012, 3.1918], device='cuda:1'), covar=tensor([0.1313, 0.1465, 0.0150, 0.0325, 0.0286, 0.0141, 0.0415, 0.0166], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0127, 0.0167, 0.0145, 0.0139, 0.0122, 0.0120], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 04:16:56,431 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96041.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:58,085 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:59,769 INFO [train.py:873] (1/4) Epoch 13, batch 5300, loss[loss=0.08813, simple_loss=0.1247, pruned_loss=0.02578, over 13912.00 frames. ], tot_loss[loss=0.1191, simple_loss=0.1513, pruned_loss=0.04347, over 1912001.24 frames. ], batch size: 20, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:17:22,099 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7623, 1.7528, 1.7093, 1.9223, 1.8521, 1.6282, 1.5183, 1.1499], device='cuda:1'), covar=tensor([0.0283, 0.0466, 0.0401, 0.0258, 0.0517, 0.0454, 0.0408, 0.0707], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0023, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:17:23,907 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.223e+02 2.915e+02 3.617e+02 9.982e+02, threshold=5.829e+02, percent-clipped=6.0 2022-12-08 04:17:25,346 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2022-12-08 04:17:30,022 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:18:03,332 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.40 vs. limit=5.0 2022-12-08 04:18:23,891 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96141.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:18:26,366 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0587, 2.9228, 2.4723, 2.6496, 3.0200, 3.0012, 3.0822, 3.0139], device='cuda:1'), covar=tensor([0.1344, 0.0973, 0.3126, 0.3632, 0.1167, 0.1417, 0.1615, 0.1310], device='cuda:1'), in_proj_covar=tensor([0.0378, 0.0269, 0.0443, 0.0557, 0.0333, 0.0430, 0.0385, 0.0372], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:18:27,058 INFO [train.py:873] (1/4) Epoch 13, batch 5400, loss[loss=0.1433, simple_loss=0.1518, pruned_loss=0.06736, over 5975.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.1523, pruned_loss=0.04375, over 1973745.35 frames. ], batch size: 100, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:18:40,881 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2542, 3.0640, 2.9853, 3.2418, 3.1288, 3.2147, 3.3417, 2.7245], device='cuda:1'), covar=tensor([0.0554, 0.0976, 0.0568, 0.0637, 0.0777, 0.0440, 0.0539, 0.0648], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0272, 0.0190, 0.0188, 0.0185, 0.0151, 0.0279, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:18:44,246 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:18:46,901 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96168.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:18:51,339 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.106e+02 2.759e+02 3.696e+02 7.492e+02, threshold=5.517e+02, percent-clipped=3.0 2022-12-08 04:19:11,233 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0430, 2.0757, 1.9446, 2.1811, 1.7838, 1.9935, 2.1177, 2.0379], device='cuda:1'), covar=tensor([0.0946, 0.1292, 0.1259, 0.0898, 0.1651, 0.0961, 0.1210, 0.0998], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0135, 0.0144, 0.0155, 0.0143, 0.0119, 0.0163, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:19:17,374 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96203.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 04:19:37,230 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96226.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:19:53,396 INFO [train.py:873] (1/4) Epoch 13, batch 5500, loss[loss=0.1246, simple_loss=0.1279, pruned_loss=0.06069, over 1327.00 frames. ], tot_loss[loss=0.12, simple_loss=0.1521, pruned_loss=0.04391, over 1933009.43 frames. ], batch size: 100, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:20:07,115 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1221, 2.0283, 2.1006, 2.1629, 2.1214, 2.0387, 2.2430, 1.8469], device='cuda:1'), covar=tensor([0.1143, 0.1385, 0.0841, 0.0850, 0.1011, 0.0812, 0.0866, 0.0832], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0273, 0.0191, 0.0189, 0.0185, 0.0152, 0.0280, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:20:17,918 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.945e+01 2.434e+02 2.937e+02 3.811e+02 8.980e+02, threshold=5.874e+02, percent-clipped=5.0 2022-12-08 04:20:18,423 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:00,279 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:02,868 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96324.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:04,191 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-08 04:21:13,571 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:21,089 INFO [train.py:873] (1/4) Epoch 13, batch 5600, loss[loss=0.1372, simple_loss=0.1609, pruned_loss=0.05674, over 14164.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1529, pruned_loss=0.04495, over 1932600.70 frames. ], batch size: 99, lr: 6.02e-03, grad_scale: 8.0 2022-12-08 04:21:39,553 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5932, 2.3007, 4.5862, 3.0957, 4.4330, 2.1667, 3.4298, 4.4098], device='cuda:1'), covar=tensor([0.0511, 0.4056, 0.0376, 0.6208, 0.0435, 0.3144, 0.1250, 0.0392], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0209, 0.0208, 0.0284, 0.0226, 0.0211, 0.0209, 0.0211], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:21:45,013 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7924, 1.7994, 1.5560, 1.9850, 1.8599, 1.7785, 1.7954, 1.6578], device='cuda:1'), covar=tensor([0.1139, 0.0886, 0.2251, 0.0623, 0.0933, 0.0587, 0.1210, 0.0746], device='cuda:1'), in_proj_covar=tensor([0.0276, 0.0295, 0.0269, 0.0271, 0.0318, 0.0300, 0.0259, 0.0248], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:21:45,613 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.301e+02 2.704e+02 3.458e+02 6.924e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-08 04:21:56,158 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:00,461 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:31,536 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 04:22:41,421 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96436.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:48,795 INFO [train.py:873] (1/4) Epoch 13, batch 5700, loss[loss=0.1852, simple_loss=0.1598, pruned_loss=0.1053, over 1253.00 frames. ], tot_loss[loss=0.1215, simple_loss=0.1531, pruned_loss=0.04501, over 1961120.93 frames. ], batch size: 100, lr: 6.02e-03, grad_scale: 8.0 2022-12-08 04:22:54,719 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96451.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:23:09,299 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96468.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:23:13,579 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.382e+02 2.762e+02 3.323e+02 6.415e+02, threshold=5.524e+02, percent-clipped=3.0 2022-12-08 04:23:40,523 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96503.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:23:51,382 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96516.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:23:55,908 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96521.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:24:10,489 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0718, 1.0427, 1.1006, 1.0398, 1.0000, 0.7533, 1.0234, 0.8462], device='cuda:1'), covar=tensor([0.0247, 0.0203, 0.0198, 0.0231, 0.0225, 0.0430, 0.0250, 0.0448], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0023, 0.0028], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:24:17,060 INFO [train.py:873] (1/4) Epoch 13, batch 5800, loss[loss=0.18, simple_loss=0.1905, pruned_loss=0.08476, over 8614.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.1523, pruned_loss=0.04413, over 1965856.02 frames. ], batch size: 100, lr: 6.02e-03, grad_scale: 4.0 2022-12-08 04:24:22,211 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:24:25,811 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2022-12-08 04:24:42,626 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.152e+02 2.229e+02 2.647e+02 3.440e+02 6.528e+02, threshold=5.293e+02, percent-clipped=3.0 2022-12-08 04:25:37,127 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96636.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:25:44,388 INFO [train.py:873] (1/4) Epoch 13, batch 5900, loss[loss=0.1186, simple_loss=0.1468, pruned_loss=0.04522, over 5979.00 frames. ], tot_loss[loss=0.1195, simple_loss=0.1518, pruned_loss=0.04359, over 2024421.23 frames. ], batch size: 100, lr: 6.01e-03, grad_scale: 4.0 2022-12-08 04:26:10,182 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 2.158e+02 2.566e+02 3.278e+02 1.082e+03, threshold=5.131e+02, percent-clipped=5.0 2022-12-08 04:26:15,613 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:26:18,896 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96684.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:01,789 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 04:27:03,987 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:11,548 INFO [train.py:873] (1/4) Epoch 13, batch 6000, loss[loss=0.1307, simple_loss=0.1582, pruned_loss=0.05158, over 6969.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.152, pruned_loss=0.04424, over 1961586.10 frames. ], batch size: 100, lr: 6.01e-03, grad_scale: 8.0 2022-12-08 04:27:11,548 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 04:27:15,808 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0796, 3.8554, 2.8520, 4.3592, 4.0643, 4.2080, 3.6783, 3.2192], device='cuda:1'), covar=tensor([0.0930, 0.1029, 0.3508, 0.0440, 0.0838, 0.1483, 0.1199, 0.2901], device='cuda:1'), in_proj_covar=tensor([0.0271, 0.0290, 0.0265, 0.0266, 0.0311, 0.0295, 0.0255, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:27:18,272 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0764, 2.4209, 3.1783, 2.8702, 2.9576, 2.0850, 2.5716, 3.0304], device='cuda:1'), covar=tensor([0.0800, 0.1408, 0.0223, 0.1121, 0.0671, 0.1023, 0.0423, 0.0330], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0208, 0.0210, 0.0285, 0.0226, 0.0212, 0.0210, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:27:19,806 INFO [train.py:905] (1/4) Epoch 13, validation: loss=0.132, simple_loss=0.1717, pruned_loss=0.04613, over 857387.00 frames. 2022-12-08 04:27:19,806 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 04:27:20,813 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96746.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:45,010 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.016e+02 2.616e+02 3.028e+02 5.890e+02, threshold=5.232e+02, percent-clipped=1.0 2022-12-08 04:27:46,844 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96776.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:53,301 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:17,515 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.48 vs. limit=5.0 2022-12-08 04:28:25,770 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:35,914 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2006, 4.4059, 4.6012, 3.9846, 4.4122, 4.6073, 1.8114, 4.1652], device='cuda:1'), covar=tensor([0.0270, 0.0289, 0.0330, 0.0493, 0.0303, 0.0201, 0.3169, 0.0264], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0172, 0.0144, 0.0141, 0.0202, 0.0136, 0.0159, 0.0190], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 04:28:40,179 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96837.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:47,194 INFO [train.py:873] (1/4) Epoch 13, batch 6100, loss[loss=0.1673, simple_loss=0.1847, pruned_loss=0.07502, over 9533.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.1522, pruned_loss=0.04445, over 1964489.10 frames. ], batch size: 100, lr: 6.01e-03, grad_scale: 8.0 2022-12-08 04:29:01,464 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7769, 1.9134, 2.6377, 2.1184, 2.6776, 2.6401, 2.4663, 2.3833], device='cuda:1'), covar=tensor([0.0722, 0.2691, 0.1144, 0.1716, 0.0557, 0.0970, 0.0797, 0.1475], device='cuda:1'), in_proj_covar=tensor([0.0357, 0.0319, 0.0403, 0.0308, 0.0381, 0.0326, 0.0369, 0.0311], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:29:07,684 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96869.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:29:12,500 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.137e+02 2.568e+02 3.326e+02 6.550e+02, threshold=5.137e+02, percent-clipped=4.0 2022-12-08 04:29:46,385 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 04:30:13,974 INFO [train.py:873] (1/4) Epoch 13, batch 6200, loss[loss=0.1291, simple_loss=0.165, pruned_loss=0.04658, over 14546.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.1522, pruned_loss=0.04439, over 1919667.68 frames. ], batch size: 43, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:30:39,951 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.243e+02 2.655e+02 3.164e+02 5.951e+02, threshold=5.310e+02, percent-clipped=2.0 2022-12-08 04:30:45,312 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:02,397 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8247, 4.5222, 4.1536, 4.4511, 4.5195, 4.7098, 4.8262, 4.7681], device='cuda:1'), covar=tensor([0.0717, 0.0409, 0.1930, 0.2494, 0.0724, 0.0789, 0.0754, 0.0757], device='cuda:1'), in_proj_covar=tensor([0.0374, 0.0264, 0.0438, 0.0551, 0.0328, 0.0427, 0.0379, 0.0370], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:31:03,556 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97000.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:28,037 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97028.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:43,226 INFO [train.py:873] (1/4) Epoch 13, batch 6300, loss[loss=0.1256, simple_loss=0.1582, pruned_loss=0.04648, over 14071.00 frames. ], tot_loss[loss=0.1204, simple_loss=0.152, pruned_loss=0.04439, over 1917484.00 frames. ], batch size: 29, lr: 6.00e-03, grad_scale: 4.0 2022-12-08 04:31:44,299 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:56,952 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97061.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:32:02,150 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2560, 1.5888, 1.7343, 1.6763, 1.5860, 1.6241, 1.4053, 1.2572], device='cuda:1'), covar=tensor([0.1646, 0.1161, 0.0372, 0.0588, 0.1165, 0.0952, 0.1590, 0.1753], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0085, 0.0066, 0.0070, 0.0098, 0.0083, 0.0098, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 04:32:09,641 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.227e+02 2.717e+02 3.666e+02 6.842e+02, threshold=5.435e+02, percent-clipped=6.0 2022-12-08 04:32:25,896 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97094.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:32:58,836 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:33:09,960 INFO [train.py:873] (1/4) Epoch 13, batch 6400, loss[loss=0.1312, simple_loss=0.1463, pruned_loss=0.05806, over 4936.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.1514, pruned_loss=0.04418, over 1890316.31 frames. ], batch size: 100, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:33:26,408 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 04:33:36,265 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.249e+02 2.635e+02 3.271e+02 6.214e+02, threshold=5.270e+02, percent-clipped=1.0 2022-12-08 04:34:09,452 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4331, 3.8488, 3.7205, 3.5953, 2.7732, 3.7269, 3.5597, 2.1778], device='cuda:1'), covar=tensor([0.1655, 0.0960, 0.0949, 0.0849, 0.0942, 0.0371, 0.1129, 0.1878], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0084, 0.0065, 0.0068, 0.0097, 0.0082, 0.0097, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 04:34:37,202 INFO [train.py:873] (1/4) Epoch 13, batch 6500, loss[loss=0.1003, simple_loss=0.1398, pruned_loss=0.03045, over 13941.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1525, pruned_loss=0.04458, over 1936083.66 frames. ], batch size: 23, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:34:50,025 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:35:03,585 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.327e+02 2.807e+02 3.571e+02 7.532e+02, threshold=5.613e+02, percent-clipped=5.0 2022-12-08 04:35:44,078 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:35:54,564 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3735, 1.6406, 2.6262, 2.0751, 2.4643, 1.7467, 2.0437, 2.3955], device='cuda:1'), covar=tensor([0.1884, 0.4268, 0.0874, 0.3394, 0.1074, 0.2974, 0.1421, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0211, 0.0214, 0.0290, 0.0230, 0.0212, 0.0213, 0.0217], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0006, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:36:02,163 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:04,578 INFO [train.py:873] (1/4) Epoch 13, batch 6600, loss[loss=0.1173, simple_loss=0.1527, pruned_loss=0.04097, over 14468.00 frames. ], tot_loss[loss=0.1197, simple_loss=0.1516, pruned_loss=0.04395, over 1949494.66 frames. ], batch size: 51, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:36:06,439 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:14,719 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:30,362 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9644, 2.0300, 2.2166, 1.4547, 1.5746, 2.0729, 1.3096, 2.0420], device='cuda:1'), covar=tensor([0.1092, 0.1372, 0.0828, 0.1774, 0.2455, 0.0911, 0.3524, 0.0993], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0099, 0.0116, 0.0087, 0.0123, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 04:36:30,384 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:31,019 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.194e+02 2.777e+02 3.424e+02 5.452e+02, threshold=5.554e+02, percent-clipped=0.0 2022-12-08 04:36:35,417 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5130, 2.5426, 2.5517, 2.6708, 2.5161, 2.3873, 1.3549, 2.2996], device='cuda:1'), covar=tensor([0.0739, 0.0679, 0.0754, 0.0522, 0.0666, 0.1321, 0.3634, 0.0637], device='cuda:1'), in_proj_covar=tensor([0.0164, 0.0170, 0.0142, 0.0139, 0.0199, 0.0135, 0.0157, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 04:36:55,836 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97403.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:37:00,395 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:20,919 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:23,867 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:32,757 INFO [train.py:873] (1/4) Epoch 13, batch 6700, loss[loss=0.117, simple_loss=0.147, pruned_loss=0.04347, over 14313.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.152, pruned_loss=0.04395, over 2016471.21 frames. ], batch size: 46, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:37:59,104 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.128e+02 2.579e+02 3.037e+02 5.604e+02, threshold=5.159e+02, percent-clipped=1.0 2022-12-08 04:38:03,381 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97480.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:38:59,022 INFO [train.py:873] (1/4) Epoch 13, batch 6800, loss[loss=0.1931, simple_loss=0.1663, pruned_loss=0.11, over 1333.00 frames. ], tot_loss[loss=0.1202, simple_loss=0.1524, pruned_loss=0.04402, over 2077575.21 frames. ], batch size: 100, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:39:08,537 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:39:25,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.339e+02 3.043e+02 3.694e+02 1.051e+03, threshold=6.086e+02, percent-clipped=5.0 2022-12-08 04:39:36,986 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 04:40:01,861 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:01,940 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:27,572 INFO [train.py:873] (1/4) Epoch 13, batch 6900, loss[loss=0.1268, simple_loss=0.1554, pruned_loss=0.04907, over 14281.00 frames. ], tot_loss[loss=0.1202, simple_loss=0.1522, pruned_loss=0.04414, over 2025287.85 frames. ], batch size: 60, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:40:36,896 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97656.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:52,689 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97674.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:40:53,028 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-08 04:40:53,268 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.182e+02 2.726e+02 3.211e+02 6.785e+02, threshold=5.451e+02, percent-clipped=1.0 2022-12-08 04:41:07,632 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0203, 2.1261, 1.9682, 2.1877, 1.7944, 2.0383, 2.1116, 2.0658], device='cuda:1'), covar=tensor([0.0897, 0.1063, 0.1019, 0.0809, 0.1479, 0.0751, 0.1040, 0.0849], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0134, 0.0143, 0.0152, 0.0141, 0.0117, 0.0161, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:41:13,549 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97698.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:41:17,871 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97703.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:18,592 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:41,326 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97730.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:45,560 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97735.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 04:41:53,810 INFO [train.py:873] (1/4) Epoch 13, batch 7000, loss[loss=0.1221, simple_loss=0.1626, pruned_loss=0.04079, over 14197.00 frames. ], tot_loss[loss=0.1197, simple_loss=0.1517, pruned_loss=0.04388, over 1939484.38 frames. ], batch size: 80, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:42:13,663 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2022-12-08 04:42:20,482 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.837e+01 2.234e+02 2.676e+02 3.367e+02 8.383e+02, threshold=5.352e+02, percent-clipped=3.0 2022-12-08 04:43:23,309 INFO [train.py:873] (1/4) Epoch 13, batch 7100, loss[loss=0.1399, simple_loss=0.1372, pruned_loss=0.07132, over 1306.00 frames. ], tot_loss[loss=0.1191, simple_loss=0.1513, pruned_loss=0.04341, over 1980803.04 frames. ], batch size: 100, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:43:49,281 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.322e+02 2.890e+02 3.385e+02 9.839e+02, threshold=5.780e+02, percent-clipped=7.0 2022-12-08 04:44:01,054 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:21,128 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:26,054 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:34,975 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-08 04:44:51,439 INFO [train.py:873] (1/4) Epoch 13, batch 7200, loss[loss=0.09436, simple_loss=0.1368, pruned_loss=0.02594, over 14369.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1526, pruned_loss=0.04448, over 1963400.59 frames. ], batch size: 41, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:44:54,891 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97949.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:45:06,842 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8196, 5.3237, 5.2533, 5.7965, 5.3862, 4.8310, 5.7022, 4.6619], device='cuda:1'), covar=tensor([0.0298, 0.0940, 0.0307, 0.0356, 0.0754, 0.0329, 0.0453, 0.0485], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0272, 0.0193, 0.0190, 0.0187, 0.0155, 0.0277, 0.0167], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 04:45:08,596 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97964.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:45:09,336 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0651, 2.1009, 1.9618, 2.1966, 1.8272, 2.0078, 2.1591, 2.0867], device='cuda:1'), covar=tensor([0.0824, 0.1082, 0.0999, 0.0800, 0.1226, 0.0869, 0.0934, 0.0837], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0135, 0.0143, 0.0152, 0.0141, 0.0117, 0.0160, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:45:18,186 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.164e+02 2.816e+02 3.534e+02 1.223e+03, threshold=5.633e+02, percent-clipped=3.0 2022-12-08 04:45:29,961 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-08 04:45:38,842 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:45:43,540 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:02,938 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8607, 1.8358, 1.9335, 1.7756, 1.7997, 1.4938, 1.6574, 1.1670], device='cuda:1'), covar=tensor([0.0166, 0.0398, 0.0274, 0.0258, 0.0200, 0.0299, 0.0289, 0.0473], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0017, 0.0018, 0.0018, 0.0029, 0.0024, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:46:07,207 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98030.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:46:07,240 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98030.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:20,635 INFO [train.py:873] (1/4) Epoch 13, batch 7300, loss[loss=0.1101, simple_loss=0.1455, pruned_loss=0.03731, over 14292.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1523, pruned_loss=0.04466, over 1931325.67 frames. ], batch size: 76, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:46:21,877 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:26,093 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:46,648 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.194e+02 2.668e+02 3.260e+02 4.997e+02, threshold=5.336e+02, percent-clipped=0.0 2022-12-08 04:46:50,049 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:47:14,610 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:47:48,797 INFO [train.py:873] (1/4) Epoch 13, batch 7400, loss[loss=0.09568, simple_loss=0.1388, pruned_loss=0.0263, over 14224.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.1519, pruned_loss=0.04391, over 1945550.13 frames. ], batch size: 35, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:47:52,965 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8778, 1.6349, 2.0957, 1.7390, 1.9978, 1.4874, 1.7236, 2.0166], device='cuda:1'), covar=tensor([0.2406, 0.2033, 0.0450, 0.1332, 0.1002, 0.1181, 0.0932, 0.0722], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0209, 0.0210, 0.0284, 0.0228, 0.0210, 0.0207, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:48:08,702 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98167.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:48:15,367 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 1.931e+02 2.491e+02 3.118e+02 6.123e+02, threshold=4.982e+02, percent-clipped=1.0 2022-12-08 04:48:44,206 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 04:48:46,002 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2022-12-08 04:48:48,146 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98211.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:07,344 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 04:49:09,356 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 04:49:16,644 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98244.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:17,647 INFO [train.py:873] (1/4) Epoch 13, batch 7500, loss[loss=0.117, simple_loss=0.1574, pruned_loss=0.03827, over 13909.00 frames. ], tot_loss[loss=0.1196, simple_loss=0.1517, pruned_loss=0.04377, over 1969104.63 frames. ], batch size: 20, lr: 5.96e-03, grad_scale: 4.0 2022-12-08 04:49:29,607 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:44,017 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.297e+02 2.932e+02 3.500e+02 6.688e+02, threshold=5.864e+02, percent-clipped=4.0 2022-12-08 04:49:46,419 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8101, 2.7307, 2.6583, 2.8894, 2.5363, 2.5651, 2.8997, 2.7999], device='cuda:1'), covar=tensor([0.0728, 0.1058, 0.0930, 0.0735, 0.1186, 0.0807, 0.0794, 0.0800], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0136, 0.0143, 0.0153, 0.0143, 0.0119, 0.0162, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:49:57,354 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7924, 1.5644, 1.9090, 2.0232, 1.4614, 1.7742, 1.8175, 1.8958], device='cuda:1'), covar=tensor([0.0180, 0.0254, 0.0160, 0.0135, 0.0265, 0.0293, 0.0192, 0.0140], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0254, 0.0367, 0.0324, 0.0265, 0.0298, 0.0303, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:50:00,705 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3666, 1.0379, 1.2739, 0.8460, 1.0719, 1.3746, 1.0008, 1.1409], device='cuda:1'), covar=tensor([0.0428, 0.1018, 0.0682, 0.0515, 0.1199, 0.0743, 0.0622, 0.1210], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0177, 0.0138, 0.0126, 0.0138, 0.0148, 0.0125, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 04:50:44,946 INFO [train.py:873] (1/4) Epoch 14, batch 0, loss[loss=0.1351, simple_loss=0.1795, pruned_loss=0.04536, over 14268.00 frames. ], tot_loss[loss=0.1351, simple_loss=0.1795, pruned_loss=0.04536, over 14268.00 frames. ], batch size: 76, lr: 5.75e-03, grad_scale: 8.0 2022-12-08 04:50:44,946 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 04:50:49,679 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9131, 4.9842, 5.1915, 4.4531, 4.9595, 5.1973, 2.1116, 4.7499], device='cuda:1'), covar=tensor([0.0132, 0.0194, 0.0202, 0.0337, 0.0196, 0.0081, 0.2885, 0.0212], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0170, 0.0142, 0.0140, 0.0201, 0.0135, 0.0158, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 04:50:52,245 INFO [train.py:905] (1/4) Epoch 14, validation: loss=0.1389, simple_loss=0.1805, pruned_loss=0.04866, over 857387.00 frames. 2022-12-08 04:50:52,246 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 04:51:01,199 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5118, 2.2908, 3.5266, 3.6385, 3.4949, 2.3240, 3.5685, 2.7420], device='cuda:1'), covar=tensor([0.0416, 0.1018, 0.0728, 0.0530, 0.0426, 0.1444, 0.0389, 0.0920], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0254, 0.0368, 0.0324, 0.0265, 0.0298, 0.0303, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:51:09,192 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8091, 1.7164, 1.7799, 1.7896, 1.7678, 1.6016, 1.7507, 1.0933], device='cuda:1'), covar=tensor([0.0229, 0.0506, 0.0341, 0.0290, 0.0225, 0.0346, 0.0300, 0.0645], device='cuda:1'), in_proj_covar=tensor([0.0018, 0.0019, 0.0017, 0.0017, 0.0018, 0.0029, 0.0024, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 04:51:11,720 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.15 vs. limit=5.0 2022-12-08 04:51:13,026 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98330.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:51:31,645 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2022-12-08 04:51:53,529 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.725e+01 2.112e+02 3.078e+02 4.479e+02 1.243e+03, threshold=6.157e+02, percent-clipped=15.0 2022-12-08 04:51:55,471 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98378.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:51:55,682 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6242, 2.6216, 4.4667, 4.5218, 4.6179, 2.7964, 4.5642, 3.6390], device='cuda:1'), covar=tensor([0.0275, 0.0972, 0.0883, 0.0358, 0.0299, 0.1313, 0.0331, 0.0702], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0255, 0.0369, 0.0326, 0.0267, 0.0299, 0.0305, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 04:52:20,898 INFO [train.py:873] (1/4) Epoch 14, batch 100, loss[loss=0.1051, simple_loss=0.146, pruned_loss=0.03205, over 14314.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.154, pruned_loss=0.04437, over 914977.78 frames. ], batch size: 28, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:52:24,292 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98410.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:53:09,629 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0001, 2.2434, 2.3044, 2.4373, 1.9915, 2.3460, 2.1266, 1.3531], device='cuda:1'), covar=tensor([0.0938, 0.1017, 0.0650, 0.0530, 0.1061, 0.0656, 0.1127, 0.2062], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0084, 0.0065, 0.0069, 0.0096, 0.0081, 0.0097, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 04:53:10,351 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:53:11,748 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 04:53:18,367 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:53:22,339 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.180e+02 2.038e+02 2.657e+02 3.395e+02 7.955e+02, threshold=5.314e+02, percent-clipped=1.0 2022-12-08 04:53:49,904 INFO [train.py:873] (1/4) Epoch 14, batch 200, loss[loss=0.09653, simple_loss=0.1374, pruned_loss=0.02785, over 14184.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1518, pruned_loss=0.04309, over 1366945.04 frames. ], batch size: 37, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:54:22,186 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98543.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:54:22,952 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98544.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:54:50,565 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.243e+02 2.774e+02 3.161e+02 6.652e+02, threshold=5.548e+02, percent-clipped=3.0 2022-12-08 04:55:03,803 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98592.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:55:14,581 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98604.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:55:16,904 INFO [train.py:873] (1/4) Epoch 14, batch 300, loss[loss=0.1224, simple_loss=0.1575, pruned_loss=0.04369, over 14553.00 frames. ], tot_loss[loss=0.1197, simple_loss=0.1516, pruned_loss=0.04388, over 1587103.85 frames. ], batch size: 43, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:56:18,355 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.234e+02 2.680e+02 3.385e+02 6.957e+02, threshold=5.360e+02, percent-clipped=3.0 2022-12-08 04:56:45,717 INFO [train.py:873] (1/4) Epoch 14, batch 400, loss[loss=0.117, simple_loss=0.1448, pruned_loss=0.04465, over 14634.00 frames. ], tot_loss[loss=0.12, simple_loss=0.1515, pruned_loss=0.04424, over 1655258.45 frames. ], batch size: 33, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 04:57:27,314 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 04:57:33,677 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98762.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:57:36,939 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:57:36,961 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0712, 3.1438, 3.2784, 3.1146, 3.2226, 2.7151, 1.4296, 3.0029], device='cuda:1'), covar=tensor([0.0374, 0.0403, 0.0391, 0.0459, 0.0351, 0.1513, 0.3083, 0.0336], device='cuda:1'), in_proj_covar=tensor([0.0163, 0.0169, 0.0141, 0.0138, 0.0198, 0.0135, 0.0157, 0.0188], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 04:57:46,469 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.221e+02 2.679e+02 3.803e+02 8.740e+02, threshold=5.359e+02, percent-clipped=4.0 2022-12-08 04:57:58,034 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1847, 3.9339, 3.6168, 3.8155, 4.0136, 4.0865, 4.1521, 4.1164], device='cuda:1'), covar=tensor([0.0818, 0.0546, 0.1924, 0.2507, 0.0772, 0.0788, 0.0867, 0.0859], device='cuda:1'), in_proj_covar=tensor([0.0380, 0.0269, 0.0442, 0.0559, 0.0337, 0.0437, 0.0390, 0.0378], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 04:58:13,695 INFO [train.py:873] (1/4) Epoch 14, batch 500, loss[loss=0.1749, simple_loss=0.1611, pruned_loss=0.09432, over 1273.00 frames. ], tot_loss[loss=0.1189, simple_loss=0.1512, pruned_loss=0.04328, over 1837274.81 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 04:58:16,982 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98810.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:14,755 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 1.935e+02 2.381e+02 2.894e+02 6.356e+02, threshold=4.763e+02, percent-clipped=3.0 2022-12-08 04:59:26,748 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98890.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:31,772 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98895.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:35,154 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:42,007 INFO [train.py:873] (1/4) Epoch 14, batch 600, loss[loss=0.1344, simple_loss=0.1346, pruned_loss=0.0671, over 2653.00 frames. ], tot_loss[loss=0.1183, simple_loss=0.1511, pruned_loss=0.04276, over 1909467.45 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 04:59:45,836 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.65 vs. limit=5.0 2022-12-08 05:00:06,951 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0656, 2.0904, 3.0614, 3.1845, 3.1051, 2.1887, 3.1001, 2.4348], device='cuda:1'), covar=tensor([0.0416, 0.0950, 0.0589, 0.0403, 0.0466, 0.1310, 0.0401, 0.0857], device='cuda:1'), in_proj_covar=tensor([0.0291, 0.0256, 0.0371, 0.0328, 0.0267, 0.0301, 0.0307, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:00:20,945 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98951.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:25,444 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98956.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:38,409 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98971.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:43,054 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.204e+01 1.954e+02 2.310e+02 3.088e+02 7.406e+02, threshold=4.620e+02, percent-clipped=2.0 2022-12-08 05:01:06,309 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.1354, 5.9849, 5.6965, 6.1750, 5.7375, 5.4942, 6.1977, 6.0054], device='cuda:1'), covar=tensor([0.0485, 0.0467, 0.0576, 0.0427, 0.0638, 0.0425, 0.0454, 0.0476], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0136, 0.0144, 0.0155, 0.0144, 0.0120, 0.0163, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:01:09,651 INFO [train.py:873] (1/4) Epoch 14, batch 700, loss[loss=0.1428, simple_loss=0.152, pruned_loss=0.06681, over 3871.00 frames. ], tot_loss[loss=0.118, simple_loss=0.1508, pruned_loss=0.04259, over 1950488.03 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 05:01:32,146 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:01:48,616 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:01,418 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:09,757 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.198e+02 2.677e+02 3.490e+02 1.171e+03, threshold=5.353e+02, percent-clipped=9.0 2022-12-08 05:02:10,717 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8762, 2.7614, 2.7267, 2.8870, 2.7735, 2.8200, 3.0012, 2.4680], device='cuda:1'), covar=tensor([0.0681, 0.1126, 0.0682, 0.0698, 0.0932, 0.0578, 0.0643, 0.0740], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0272, 0.0192, 0.0190, 0.0186, 0.0153, 0.0278, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:02:37,294 INFO [train.py:873] (1/4) Epoch 14, batch 800, loss[loss=0.1174, simple_loss=0.1524, pruned_loss=0.04123, over 13954.00 frames. ], tot_loss[loss=0.1181, simple_loss=0.1506, pruned_loss=0.04287, over 1940219.53 frames. ], batch size: 23, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:02:41,546 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99112.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:43,048 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99114.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:45,925 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4945, 2.1001, 2.5153, 2.5855, 2.4419, 2.0269, 2.5852, 2.3158], device='cuda:1'), covar=tensor([0.0365, 0.0810, 0.0527, 0.0401, 0.0494, 0.1059, 0.0439, 0.0587], device='cuda:1'), in_proj_covar=tensor([0.0288, 0.0254, 0.0369, 0.0325, 0.0265, 0.0299, 0.0304, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:02:54,554 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7770, 1.1153, 1.2643, 1.2746, 1.0672, 1.3051, 1.1024, 0.9353], device='cuda:1'), covar=tensor([0.2208, 0.1081, 0.0554, 0.0501, 0.1431, 0.0935, 0.1392, 0.1411], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0084, 0.0067, 0.0070, 0.0097, 0.0083, 0.0098, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:03:35,471 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1227, 2.0663, 1.8052, 1.8928, 2.0788, 2.1176, 2.0996, 2.0624], device='cuda:1'), covar=tensor([0.1319, 0.1089, 0.3028, 0.2988, 0.1229, 0.1289, 0.1815, 0.1344], device='cuda:1'), in_proj_covar=tensor([0.0374, 0.0263, 0.0435, 0.0553, 0.0334, 0.0435, 0.0384, 0.0373], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:03:38,207 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.181e+02 2.713e+02 3.365e+02 5.447e+02, threshold=5.425e+02, percent-clipped=1.0 2022-12-08 05:03:58,135 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99199.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:05,015 INFO [train.py:873] (1/4) Epoch 14, batch 900, loss[loss=0.1601, simple_loss=0.1473, pruned_loss=0.08647, over 1225.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1509, pruned_loss=0.04349, over 1897696.99 frames. ], batch size: 100, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:04:24,544 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:39,143 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99246.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:39,941 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:43,478 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99251.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:58,900 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:05,665 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.212e+02 2.795e+02 3.543e+02 9.051e+02, threshold=5.590e+02, percent-clipped=4.0 2022-12-08 05:05:17,529 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99290.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:32,837 INFO [train.py:873] (1/4) Epoch 14, batch 1000, loss[loss=0.1369, simple_loss=0.1604, pruned_loss=0.05671, over 14330.00 frames. ], tot_loss[loss=0.1189, simple_loss=0.1511, pruned_loss=0.04338, over 1972166.34 frames. ], batch size: 55, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:05:47,389 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99324.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:49,840 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99327.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:51,626 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:06:32,477 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.112e+02 2.804e+02 3.540e+02 1.187e+03, threshold=5.607e+02, percent-clipped=4.0 2022-12-08 05:06:41,127 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:06:45,121 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3722, 4.2056, 3.7430, 3.8812, 4.2449, 4.3379, 4.4783, 4.3567], device='cuda:1'), covar=tensor([0.1102, 0.0782, 0.2476, 0.3511, 0.0938, 0.1126, 0.1123, 0.1146], device='cuda:1'), in_proj_covar=tensor([0.0378, 0.0264, 0.0438, 0.0554, 0.0334, 0.0437, 0.0386, 0.0374], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:06:59,881 INFO [train.py:873] (1/4) Epoch 14, batch 1100, loss[loss=0.1374, simple_loss=0.1431, pruned_loss=0.06587, over 2602.00 frames. ], tot_loss[loss=0.1194, simple_loss=0.151, pruned_loss=0.0439, over 1938345.34 frames. ], batch size: 100, lr: 5.71e-03, grad_scale: 8.0 2022-12-08 05:06:59,971 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:07:12,222 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:07:12,251 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8493, 1.8589, 2.9965, 2.2441, 2.8257, 1.8921, 2.4528, 2.8689], device='cuda:1'), covar=tensor([0.1110, 0.3881, 0.0693, 0.4031, 0.1015, 0.2789, 0.1132, 0.0659], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0208, 0.0212, 0.0281, 0.0228, 0.0207, 0.0206, 0.0211], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:07:53,815 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99468.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:08:00,247 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.195e+02 2.512e+02 3.214e+02 5.194e+02, threshold=5.024e+02, percent-clipped=0.0 2022-12-08 05:08:04,783 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99481.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:08:27,690 INFO [train.py:873] (1/4) Epoch 14, batch 1200, loss[loss=0.1183, simple_loss=0.1526, pruned_loss=0.04195, over 14252.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1505, pruned_loss=0.04373, over 1963770.58 frames. ], batch size: 66, lr: 5.71e-03, grad_scale: 8.0 2022-12-08 05:08:28,445 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-08 05:08:34,718 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8590, 4.7332, 4.4738, 4.9514, 4.5107, 4.2499, 4.9322, 4.7284], device='cuda:1'), covar=tensor([0.0618, 0.0719, 0.0730, 0.0462, 0.0669, 0.0623, 0.0491, 0.0683], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0136, 0.0144, 0.0153, 0.0142, 0.0119, 0.0162, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:08:46,741 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99529.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 05:09:01,589 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99546.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:05,833 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:09,000 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1274, 3.8187, 3.5636, 3.7340, 3.9582, 4.0310, 4.0789, 4.0768], device='cuda:1'), covar=tensor([0.0785, 0.0599, 0.2130, 0.2602, 0.0781, 0.0824, 0.0924, 0.0782], device='cuda:1'), in_proj_covar=tensor([0.0380, 0.0268, 0.0441, 0.0562, 0.0339, 0.0441, 0.0389, 0.0376], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:09:18,964 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-08 05:09:20,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99568.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:27,999 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.156e+02 2.823e+02 3.405e+02 6.610e+02, threshold=5.645e+02, percent-clipped=3.0 2022-12-08 05:09:35,433 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99585.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:38,980 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99589.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:42,998 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:47,155 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99599.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:54,364 INFO [train.py:873] (1/4) Epoch 14, batch 1300, loss[loss=0.172, simple_loss=0.1526, pruned_loss=0.0957, over 1245.00 frames. ], tot_loss[loss=0.1185, simple_loss=0.1505, pruned_loss=0.04324, over 2005580.74 frames. ], batch size: 100, lr: 5.71e-03, grad_scale: 4.0 2022-12-08 05:10:10,441 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:12,924 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99627.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:14,842 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99629.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:32,858 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:54,759 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99675.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:56,553 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.185e+02 2.869e+02 3.573e+02 7.828e+02, threshold=5.739e+02, percent-clipped=4.0 2022-12-08 05:10:59,275 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:59,358 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:11:22,228 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-08 05:11:23,268 INFO [train.py:873] (1/4) Epoch 14, batch 1400, loss[loss=0.08842, simple_loss=0.1392, pruned_loss=0.01882, over 14252.00 frames. ], tot_loss[loss=0.1181, simple_loss=0.1505, pruned_loss=0.04289, over 1962612.09 frames. ], batch size: 35, lr: 5.71e-03, grad_scale: 4.0 2022-12-08 05:11:23,378 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:11:52,861 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99741.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:11:56,946 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2022-12-08 05:12:05,056 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99755.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:12:23,740 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99776.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:12:24,429 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.204e+02 2.049e+02 2.625e+02 3.154e+02 6.097e+02, threshold=5.251e+02, percent-clipped=2.0 2022-12-08 05:12:50,315 INFO [train.py:873] (1/4) Epoch 14, batch 1500, loss[loss=0.1215, simple_loss=0.1535, pruned_loss=0.04477, over 14531.00 frames. ], tot_loss[loss=0.1173, simple_loss=0.1499, pruned_loss=0.04234, over 1974580.88 frames. ], batch size: 43, lr: 5.70e-03, grad_scale: 4.0 2022-12-08 05:13:06,038 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99824.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:13:39,849 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 05:13:51,974 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.709e+01 2.163e+02 2.561e+02 3.336e+02 7.134e+02, threshold=5.122e+02, percent-clipped=2.0 2022-12-08 05:13:58,890 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99885.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:01,370 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6323, 2.5031, 2.2705, 2.3868, 2.5602, 2.5657, 2.5502, 2.6092], device='cuda:1'), covar=tensor([0.1065, 0.0766, 0.2406, 0.2768, 0.1044, 0.1213, 0.1538, 0.0907], device='cuda:1'), in_proj_covar=tensor([0.0379, 0.0265, 0.0436, 0.0556, 0.0339, 0.0438, 0.0388, 0.0374], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:14:03,361 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7800, 2.0375, 2.6626, 2.2422, 2.7236, 2.5858, 2.4985, 2.3211], device='cuda:1'), covar=tensor([0.0786, 0.2745, 0.0864, 0.1767, 0.0716, 0.1009, 0.1068, 0.1731], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0316, 0.0400, 0.0306, 0.0379, 0.0323, 0.0365, 0.0307], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:14:18,359 INFO [train.py:873] (1/4) Epoch 14, batch 1600, loss[loss=0.1596, simple_loss=0.1431, pruned_loss=0.08806, over 1228.00 frames. ], tot_loss[loss=0.1177, simple_loss=0.1502, pruned_loss=0.04257, over 1988056.69 frames. ], batch size: 100, lr: 5.70e-03, grad_scale: 8.0 2022-12-08 05:14:32,776 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:32,827 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:40,492 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:45,275 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1669, 2.0366, 2.0949, 2.1984, 2.1112, 2.1171, 2.2356, 1.9310], device='cuda:1'), covar=tensor([0.0875, 0.1239, 0.0839, 0.0733, 0.0977, 0.0698, 0.0888, 0.0750], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0272, 0.0192, 0.0192, 0.0184, 0.0155, 0.0281, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:14:52,038 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:15:15,682 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:15:20,185 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.180e+02 2.741e+02 3.315e+02 6.557e+02, threshold=5.482e+02, percent-clipped=1.0 2022-12-08 05:15:23,156 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:15:25,749 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0981, 2.0520, 2.4056, 1.4540, 1.6041, 2.2067, 1.3355, 2.1130], device='cuda:1'), covar=tensor([0.1055, 0.1730, 0.0715, 0.2274, 0.2498, 0.0908, 0.3181, 0.1146], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0099, 0.0090, 0.0097, 0.0114, 0.0087, 0.0122, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 05:15:28,175 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3036, 2.3219, 4.3147, 2.9696, 4.1335, 1.9431, 3.1676, 4.0214], device='cuda:1'), covar=tensor([0.0562, 0.3877, 0.0492, 0.5727, 0.0529, 0.3561, 0.1324, 0.0516], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0207, 0.0210, 0.0281, 0.0227, 0.0208, 0.0206, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:15:35,011 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3445, 0.9993, 1.2813, 0.8906, 1.0807, 1.3804, 1.0045, 1.0270], device='cuda:1'), covar=tensor([0.0371, 0.0849, 0.0539, 0.0550, 0.0918, 0.0701, 0.0516, 0.1300], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0175, 0.0136, 0.0125, 0.0139, 0.0149, 0.0127, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 05:15:50,677 INFO [train.py:873] (1/4) Epoch 14, batch 1700, loss[loss=0.1095, simple_loss=0.1432, pruned_loss=0.03787, over 14299.00 frames. ], tot_loss[loss=0.1176, simple_loss=0.1503, pruned_loss=0.04249, over 1961268.43 frames. ], batch size: 63, lr: 5.70e-03, grad_scale: 8.0 2022-12-08 05:15:50,807 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100007.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:16:09,970 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100028.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:16,924 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:24,764 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2821, 1.7858, 2.4596, 2.0905, 2.3239, 1.7061, 2.0104, 2.2949], device='cuda:1'), covar=tensor([0.1828, 0.3312, 0.0523, 0.1985, 0.1151, 0.2063, 0.1107, 0.0844], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0207, 0.0209, 0.0280, 0.0226, 0.0206, 0.0206, 0.0211], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:16:25,734 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2022-12-08 05:16:27,950 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9884, 5.4607, 5.3278, 5.9585, 5.5092, 4.8058, 5.8575, 4.8036], device='cuda:1'), covar=tensor([0.0322, 0.0803, 0.0338, 0.0309, 0.0679, 0.0321, 0.0445, 0.0470], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0271, 0.0192, 0.0191, 0.0183, 0.0153, 0.0280, 0.0167], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:16:28,899 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9028, 1.6212, 2.0746, 1.7537, 1.9696, 1.4569, 1.6742, 1.9763], device='cuda:1'), covar=tensor([0.2146, 0.2673, 0.0483, 0.1337, 0.1080, 0.1384, 0.1138, 0.0741], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0207, 0.0209, 0.0280, 0.0226, 0.0206, 0.0207, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:16:45,312 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100068.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:16:51,760 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100076.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:52,530 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.197e+02 2.632e+02 3.389e+02 8.054e+02, threshold=5.264e+02, percent-clipped=5.0 2022-12-08 05:16:52,652 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4909, 4.3119, 4.1826, 4.5187, 4.1471, 3.8967, 4.5432, 4.3506], device='cuda:1'), covar=tensor([0.0639, 0.0828, 0.0854, 0.0638, 0.0861, 0.0671, 0.0566, 0.0790], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0136, 0.0142, 0.0155, 0.0143, 0.0119, 0.0162, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:17:19,679 INFO [train.py:873] (1/4) Epoch 14, batch 1800, loss[loss=0.079, simple_loss=0.1158, pruned_loss=0.02108, over 10776.00 frames. ], tot_loss[loss=0.1172, simple_loss=0.1501, pruned_loss=0.04217, over 1969916.46 frames. ], batch size: 13, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:17:31,521 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2022-12-08 05:17:34,615 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:17:34,680 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100124.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 05:18:17,540 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100172.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:18:21,977 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.047e+02 2.482e+02 3.140e+02 5.159e+02, threshold=4.965e+02, percent-clipped=0.0 2022-12-08 05:18:48,335 INFO [train.py:873] (1/4) Epoch 14, batch 1900, loss[loss=0.1045, simple_loss=0.1462, pruned_loss=0.03143, over 14047.00 frames. ], tot_loss[loss=0.1182, simple_loss=0.151, pruned_loss=0.04266, over 2051873.66 frames. ], batch size: 19, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:19:03,360 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100224.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:15,349 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:21,286 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100245.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:45,373 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:49,479 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.314e+02 2.818e+02 3.736e+02 1.298e+03, threshold=5.635e+02, percent-clipped=11.0 2022-12-08 05:20:03,317 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:20:08,734 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:20:09,168 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 05:20:15,746 INFO [train.py:873] (1/4) Epoch 14, batch 2000, loss[loss=0.1222, simple_loss=0.1273, pruned_loss=0.05857, over 2727.00 frames. ], tot_loss[loss=0.1183, simple_loss=0.1513, pruned_loss=0.04266, over 2107975.57 frames. ], batch size: 100, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:20:41,571 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:20:58,985 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3405, 1.9749, 2.2733, 1.4817, 1.9371, 2.2512, 2.2970, 1.9809], device='cuda:1'), covar=tensor([0.0781, 0.0681, 0.0850, 0.1526, 0.1208, 0.0700, 0.0649, 0.1433], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0174, 0.0136, 0.0125, 0.0138, 0.0149, 0.0127, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 05:21:05,112 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100363.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 05:21:17,436 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 1.945e+02 2.468e+02 3.146e+02 6.937e+02, threshold=4.935e+02, percent-clipped=2.0 2022-12-08 05:21:23,783 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:21:39,834 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9643, 4.5371, 4.4472, 4.9418, 4.6315, 4.3148, 4.9083, 4.2421], device='cuda:1'), covar=tensor([0.0336, 0.0840, 0.0342, 0.0358, 0.0679, 0.0583, 0.0463, 0.0422], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0267, 0.0188, 0.0187, 0.0179, 0.0151, 0.0274, 0.0163], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:21:39,940 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100402.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:21:44,076 INFO [train.py:873] (1/4) Epoch 14, batch 2100, loss[loss=0.1246, simple_loss=0.1495, pruned_loss=0.04986, over 5953.00 frames. ], tot_loss[loss=0.118, simple_loss=0.1507, pruned_loss=0.04264, over 2023766.62 frames. ], batch size: 100, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:21:56,612 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1384, 4.9070, 4.7555, 5.1116, 4.7296, 4.5221, 5.2027, 4.9748], device='cuda:1'), covar=tensor([0.0549, 0.0667, 0.0581, 0.0483, 0.0585, 0.0433, 0.0507, 0.0567], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0137, 0.0142, 0.0154, 0.0143, 0.0118, 0.0162, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:22:08,308 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.47 vs. limit=5.0 2022-12-08 05:22:34,511 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100463.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:22:36,748 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 05:22:46,710 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.182e+02 2.587e+02 3.120e+02 6.565e+02, threshold=5.173e+02, percent-clipped=5.0 2022-12-08 05:22:46,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4295, 2.1049, 2.2795, 1.5665, 2.0735, 2.2965, 2.4268, 2.0374], device='cuda:1'), covar=tensor([0.0826, 0.0662, 0.1011, 0.1671, 0.1065, 0.0967, 0.0608, 0.1473], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0173, 0.0136, 0.0125, 0.0138, 0.0149, 0.0126, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 05:23:06,320 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0471, 2.0033, 2.0638, 2.0900, 2.0031, 1.3569, 1.8107, 2.1111], device='cuda:1'), covar=tensor([0.1208, 0.0737, 0.1287, 0.0794, 0.1245, 0.0864, 0.0821, 0.0464], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0029, 0.0030, 0.0044, 0.0031, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:23:12,708 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5437, 2.6836, 4.2980, 3.2795, 4.3176, 4.0693, 3.9987, 3.6159], device='cuda:1'), covar=tensor([0.0740, 0.3386, 0.0871, 0.1762, 0.0780, 0.1029, 0.1879, 0.1747], device='cuda:1'), in_proj_covar=tensor([0.0360, 0.0321, 0.0403, 0.0309, 0.0385, 0.0327, 0.0367, 0.0310], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:23:13,307 INFO [train.py:873] (1/4) Epoch 14, batch 2200, loss[loss=0.1091, simple_loss=0.1488, pruned_loss=0.0347, over 14136.00 frames. ], tot_loss[loss=0.1173, simple_loss=0.1504, pruned_loss=0.04213, over 2030644.13 frames. ], batch size: 29, lr: 5.68e-03, grad_scale: 8.0 2022-12-08 05:23:51,366 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 05:24:10,082 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5623, 2.2748, 2.9566, 1.8995, 1.9021, 2.7077, 1.4507, 2.6023], device='cuda:1'), covar=tensor([0.1071, 0.1423, 0.0845, 0.2372, 0.2588, 0.0939, 0.3657, 0.1032], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0099, 0.0091, 0.0099, 0.0116, 0.0088, 0.0122, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 05:24:15,306 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 2.000e+02 2.520e+02 3.271e+02 5.030e+02, threshold=5.040e+02, percent-clipped=0.0 2022-12-08 05:24:19,727 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 05:24:30,045 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:24:41,407 INFO [train.py:873] (1/4) Epoch 14, batch 2300, loss[loss=0.1256, simple_loss=0.1476, pruned_loss=0.05179, over 5961.00 frames. ], tot_loss[loss=0.1178, simple_loss=0.1506, pruned_loss=0.0425, over 1954189.95 frames. ], batch size: 100, lr: 5.68e-03, grad_scale: 4.0 2022-12-08 05:25:07,727 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 05:25:09,315 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2022-12-08 05:25:30,678 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100663.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:25:44,076 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 1.999e+02 2.526e+02 3.011e+02 4.450e+02, threshold=5.052e+02, percent-clipped=0.0 2022-12-08 05:26:09,986 INFO [train.py:873] (1/4) Epoch 14, batch 2400, loss[loss=0.1225, simple_loss=0.1395, pruned_loss=0.05273, over 4939.00 frames. ], tot_loss[loss=0.1178, simple_loss=0.1505, pruned_loss=0.04252, over 1931855.02 frames. ], batch size: 100, lr: 5.68e-03, grad_scale: 8.0 2022-12-08 05:26:13,585 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100711.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:26:30,432 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100730.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:26:55,066 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100758.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:27:02,114 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9101, 1.5405, 3.2332, 2.9439, 3.1208, 3.2477, 2.5091, 3.2133], device='cuda:1'), covar=tensor([0.1262, 0.1368, 0.0144, 0.0324, 0.0262, 0.0134, 0.0346, 0.0168], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0157, 0.0128, 0.0166, 0.0144, 0.0138, 0.0121, 0.0121], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 05:27:13,001 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.043e+02 2.633e+02 3.365e+02 7.380e+02, threshold=5.267e+02, percent-clipped=6.0 2022-12-08 05:27:24,808 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100791.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 05:27:34,544 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.14 vs. limit=5.0 2022-12-08 05:27:38,941 INFO [train.py:873] (1/4) Epoch 14, batch 2500, loss[loss=0.1169, simple_loss=0.1541, pruned_loss=0.03987, over 14420.00 frames. ], tot_loss[loss=0.1177, simple_loss=0.1502, pruned_loss=0.04254, over 1946415.18 frames. ], batch size: 73, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:28:25,803 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4129, 1.5267, 1.4411, 1.4295, 1.5967, 0.9394, 1.3037, 1.2863], device='cuda:1'), covar=tensor([0.0557, 0.0570, 0.0758, 0.0855, 0.0531, 0.0954, 0.0741, 0.0699], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0029, 0.0030, 0.0044, 0.0031, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:28:41,717 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.248e+02 2.659e+02 3.410e+02 6.665e+02, threshold=5.318e+02, percent-clipped=2.0 2022-12-08 05:28:45,983 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9502, 1.5229, 1.9867, 1.3079, 1.6565, 2.0102, 1.8720, 1.7145], device='cuda:1'), covar=tensor([0.0827, 0.0731, 0.0974, 0.1695, 0.1642, 0.1012, 0.0747, 0.1809], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0176, 0.0137, 0.0125, 0.0138, 0.0151, 0.0127, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 05:28:55,501 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100894.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:29:06,706 INFO [train.py:873] (1/4) Epoch 14, batch 2600, loss[loss=0.1416, simple_loss=0.1692, pruned_loss=0.05704, over 9473.00 frames. ], tot_loss[loss=0.1178, simple_loss=0.1499, pruned_loss=0.04286, over 1900711.61 frames. ], batch size: 100, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:29:37,210 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100942.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:30:08,942 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.110e+02 2.658e+02 3.162e+02 7.088e+02, threshold=5.316e+02, percent-clipped=3.0 2022-12-08 05:30:34,934 INFO [train.py:873] (1/4) Epoch 14, batch 2700, loss[loss=0.1083, simple_loss=0.1461, pruned_loss=0.03522, over 14270.00 frames. ], tot_loss[loss=0.1175, simple_loss=0.15, pruned_loss=0.04244, over 1925763.84 frames. ], batch size: 76, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:30:38,477 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1302, 1.4884, 1.6702, 1.6826, 1.5524, 1.6103, 1.3263, 1.2985], device='cuda:1'), covar=tensor([0.1660, 0.1098, 0.0374, 0.0444, 0.1251, 0.0937, 0.1811, 0.1632], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0083, 0.0065, 0.0070, 0.0096, 0.0082, 0.0097, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:31:20,773 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:31:38,598 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.205e+02 2.633e+02 3.607e+02 6.784e+02, threshold=5.265e+02, percent-clipped=5.0 2022-12-08 05:31:45,715 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101086.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:32:03,783 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:32:04,577 INFO [train.py:873] (1/4) Epoch 14, batch 2800, loss[loss=0.1195, simple_loss=0.1586, pruned_loss=0.04021, over 13917.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1496, pruned_loss=0.04177, over 1966128.37 frames. ], batch size: 23, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:32:04,758 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101107.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:32:58,736 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101168.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:33:07,500 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.080e+02 2.537e+02 3.233e+02 5.412e+02, threshold=5.073e+02, percent-clipped=2.0 2022-12-08 05:33:11,937 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7289, 1.9568, 2.0739, 2.2074, 1.9535, 2.1592, 1.8688, 1.4710], device='cuda:1'), covar=tensor([0.1156, 0.1181, 0.0730, 0.0468, 0.0934, 0.0736, 0.1241, 0.2029], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0084, 0.0066, 0.0070, 0.0096, 0.0083, 0.0098, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:33:33,023 INFO [train.py:873] (1/4) Epoch 14, batch 2900, loss[loss=0.1773, simple_loss=0.1551, pruned_loss=0.09975, over 1269.00 frames. ], tot_loss[loss=0.116, simple_loss=0.1489, pruned_loss=0.04151, over 1924873.45 frames. ], batch size: 100, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:34:08,554 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8880, 1.5644, 2.0503, 1.6909, 1.9290, 1.4848, 1.6252, 1.9757], device='cuda:1'), covar=tensor([0.2142, 0.2805, 0.0603, 0.1640, 0.1161, 0.1339, 0.1263, 0.0827], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0207, 0.0212, 0.0279, 0.0230, 0.0209, 0.0207, 0.0214], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:34:15,919 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8133, 1.9161, 2.0756, 2.2866, 2.1523, 1.9129, 1.7516, 1.9678], device='cuda:1'), covar=tensor([0.0480, 0.0496, 0.0421, 0.0233, 0.0338, 0.0499, 0.0522, 0.0449], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0019, 0.0017, 0.0018, 0.0018, 0.0029, 0.0024, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:34:36,774 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.145e+02 2.615e+02 3.247e+02 7.058e+02, threshold=5.230e+02, percent-clipped=3.0 2022-12-08 05:34:59,182 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5159, 2.7567, 4.2966, 3.1675, 4.2735, 4.2256, 3.9921, 3.6761], device='cuda:1'), covar=tensor([0.0628, 0.2726, 0.0944, 0.1681, 0.0742, 0.0730, 0.1456, 0.1653], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0314, 0.0396, 0.0303, 0.0376, 0.0322, 0.0360, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:35:01,746 INFO [train.py:873] (1/4) Epoch 14, batch 3000, loss[loss=0.1075, simple_loss=0.1416, pruned_loss=0.03669, over 14406.00 frames. ], tot_loss[loss=0.1159, simple_loss=0.1491, pruned_loss=0.04132, over 1967253.89 frames. ], batch size: 53, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:35:01,747 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 05:35:09,249 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0277, 2.8696, 2.4482, 3.0297, 2.9957, 2.9256, 2.5008, 2.4203], device='cuda:1'), covar=tensor([0.0653, 0.0831, 0.2165, 0.0696, 0.0935, 0.0742, 0.1576, 0.1987], device='cuda:1'), in_proj_covar=tensor([0.0271, 0.0293, 0.0262, 0.0269, 0.0315, 0.0294, 0.0256, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:35:10,213 INFO [train.py:905] (1/4) Epoch 14, validation: loss=0.134, simple_loss=0.1722, pruned_loss=0.04793, over 857387.00 frames. 2022-12-08 05:35:10,214 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 05:35:18,321 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.70 vs. limit=5.0 2022-12-08 05:36:12,836 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.305e+02 3.036e+02 3.734e+02 6.961e+02, threshold=6.071e+02, percent-clipped=4.0 2022-12-08 05:36:18,783 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101386.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:36:37,208 INFO [train.py:873] (1/4) Epoch 14, batch 3100, loss[loss=0.1159, simple_loss=0.1464, pruned_loss=0.04271, over 14244.00 frames. ], tot_loss[loss=0.1158, simple_loss=0.1488, pruned_loss=0.04139, over 1920632.02 frames. ], batch size: 69, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:37:00,779 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101434.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:37:26,079 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101463.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:37:33,332 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9312, 2.5378, 3.8571, 4.0410, 3.8604, 2.4110, 3.8521, 2.9856], device='cuda:1'), covar=tensor([0.0335, 0.0910, 0.0713, 0.0399, 0.0393, 0.1418, 0.0399, 0.0856], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0253, 0.0370, 0.0326, 0.0265, 0.0299, 0.0301, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:37:39,841 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.043e+01 2.198e+02 2.600e+02 3.209e+02 7.473e+02, threshold=5.200e+02, percent-clipped=3.0 2022-12-08 05:37:57,327 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5989, 2.6400, 2.7571, 2.6093, 2.7079, 2.3434, 1.4367, 2.4971], device='cuda:1'), covar=tensor([0.0493, 0.0478, 0.0419, 0.0431, 0.0369, 0.1332, 0.2643, 0.0406], device='cuda:1'), in_proj_covar=tensor([0.0169, 0.0174, 0.0143, 0.0141, 0.0203, 0.0139, 0.0159, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 05:38:04,114 INFO [train.py:873] (1/4) Epoch 14, batch 3200, loss[loss=0.1308, simple_loss=0.1639, pruned_loss=0.04879, over 14507.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1493, pruned_loss=0.04173, over 1951288.24 frames. ], batch size: 49, lr: 5.65e-03, grad_scale: 8.0 2022-12-08 05:38:23,858 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 05:39:08,337 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.131e+02 2.492e+02 3.275e+02 6.158e+02, threshold=4.985e+02, percent-clipped=2.0 2022-12-08 05:39:32,250 INFO [train.py:873] (1/4) Epoch 14, batch 3300, loss[loss=0.09835, simple_loss=0.1404, pruned_loss=0.02813, over 14635.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1494, pruned_loss=0.04174, over 1978213.28 frames. ], batch size: 23, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:40:36,072 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.167e+02 2.269e+02 2.791e+02 3.629e+02 1.218e+03, threshold=5.581e+02, percent-clipped=7.0 2022-12-08 05:40:39,672 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101684.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:40:55,329 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0617, 1.8971, 4.1414, 3.8030, 3.9073, 4.1910, 3.5507, 4.1818], device='cuda:1'), covar=tensor([0.1415, 0.1391, 0.0103, 0.0225, 0.0202, 0.0122, 0.0215, 0.0122], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0158, 0.0129, 0.0168, 0.0145, 0.0140, 0.0121, 0.0122], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 05:40:59,540 INFO [train.py:873] (1/4) Epoch 14, batch 3400, loss[loss=0.157, simple_loss=0.1382, pruned_loss=0.08785, over 1300.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.1491, pruned_loss=0.04112, over 2008751.29 frames. ], batch size: 100, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:41:04,558 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 05:41:32,700 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101745.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:41:46,412 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9408, 5.3963, 5.2801, 5.8634, 5.3791, 4.9032, 5.8113, 4.7319], device='cuda:1'), covar=tensor([0.0329, 0.0940, 0.0318, 0.0366, 0.0817, 0.0361, 0.0459, 0.0488], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0273, 0.0191, 0.0189, 0.0182, 0.0153, 0.0280, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:41:48,194 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:42:02,723 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.274e+02 2.665e+02 3.242e+02 5.348e+02, threshold=5.329e+02, percent-clipped=0.0 2022-12-08 05:42:26,267 INFO [train.py:873] (1/4) Epoch 14, batch 3500, loss[loss=0.1777, simple_loss=0.1578, pruned_loss=0.09877, over 1293.00 frames. ], tot_loss[loss=0.1154, simple_loss=0.1491, pruned_loss=0.04084, over 2027176.57 frames. ], batch size: 100, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:42:29,421 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101811.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:42:36,208 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0462, 2.7593, 2.8090, 1.8446, 2.5139, 2.8531, 3.0338, 2.4834], device='cuda:1'), covar=tensor([0.0641, 0.0795, 0.0811, 0.1442, 0.1001, 0.0587, 0.0589, 0.1119], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0177, 0.0137, 0.0127, 0.0139, 0.0151, 0.0128, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:43:29,694 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.210e+02 2.811e+02 3.369e+02 8.008e+02, threshold=5.622e+02, percent-clipped=4.0 2022-12-08 05:43:53,151 INFO [train.py:873] (1/4) Epoch 14, batch 3600, loss[loss=0.1526, simple_loss=0.1463, pruned_loss=0.07942, over 1228.00 frames. ], tot_loss[loss=0.1171, simple_loss=0.1501, pruned_loss=0.04202, over 1966259.16 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:44:05,120 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101920.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:44:10,443 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9275, 2.4381, 3.9201, 4.0509, 3.9483, 2.3754, 3.9568, 3.1121], device='cuda:1'), covar=tensor([0.0375, 0.0945, 0.0777, 0.0443, 0.0388, 0.1490, 0.0431, 0.0840], device='cuda:1'), in_proj_covar=tensor([0.0287, 0.0253, 0.0368, 0.0324, 0.0265, 0.0299, 0.0301, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:44:13,681 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2022-12-08 05:44:40,783 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3913, 2.4561, 2.5635, 2.4711, 2.4984, 2.0614, 1.4777, 2.2655], device='cuda:1'), covar=tensor([0.0570, 0.0461, 0.0434, 0.0411, 0.0422, 0.1374, 0.2456, 0.0407], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0176, 0.0145, 0.0143, 0.0206, 0.0140, 0.0161, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 05:44:57,575 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.121e+02 2.665e+02 3.565e+02 6.931e+02, threshold=5.329e+02, percent-clipped=3.0 2022-12-08 05:44:58,541 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9730, 4.7150, 4.5672, 4.9964, 4.6641, 4.1658, 5.0828, 4.8818], device='cuda:1'), covar=tensor([0.0680, 0.0746, 0.0751, 0.0646, 0.0671, 0.0535, 0.0563, 0.0714], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0135, 0.0140, 0.0154, 0.0141, 0.0118, 0.0162, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 05:44:58,650 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101981.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:45:03,558 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:45:20,306 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9040, 3.6654, 3.5406, 3.9350, 3.6494, 3.6319, 3.9632, 3.2594], device='cuda:1'), covar=tensor([0.0490, 0.1003, 0.0463, 0.0441, 0.0804, 0.1168, 0.0522, 0.0571], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0270, 0.0190, 0.0188, 0.0180, 0.0153, 0.0277, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:45:21,974 INFO [train.py:873] (1/4) Epoch 14, batch 3700, loss[loss=0.1142, simple_loss=0.1399, pruned_loss=0.04419, over 6030.00 frames. ], tot_loss[loss=0.1177, simple_loss=0.1504, pruned_loss=0.04248, over 1954661.02 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:45:51,400 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:45:58,260 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102048.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:46:24,012 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2022-12-08 05:46:24,433 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1384, 3.4678, 3.3740, 3.4399, 2.5306, 3.4147, 3.2258, 1.8199], device='cuda:1'), covar=tensor([0.1035, 0.0572, 0.0717, 0.0355, 0.0913, 0.0432, 0.0907, 0.2088], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0084, 0.0066, 0.0071, 0.0096, 0.0082, 0.0098, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:46:26,400 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.149e+02 2.707e+02 3.533e+02 7.252e+02, threshold=5.415e+02, percent-clipped=4.0 2022-12-08 05:46:49,971 INFO [train.py:873] (1/4) Epoch 14, batch 3800, loss[loss=0.1267, simple_loss=0.1379, pruned_loss=0.05778, over 4945.00 frames. ], tot_loss[loss=0.117, simple_loss=0.1499, pruned_loss=0.04202, over 1945094.05 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:46:58,787 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-08 05:47:03,037 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2016, 2.9836, 2.7682, 2.8968, 3.1639, 3.1570, 3.1874, 3.1706], device='cuda:1'), covar=tensor([0.1056, 0.0715, 0.2361, 0.2799, 0.0858, 0.1003, 0.1171, 0.1006], device='cuda:1'), in_proj_covar=tensor([0.0384, 0.0268, 0.0445, 0.0567, 0.0343, 0.0439, 0.0389, 0.0381], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:47:23,223 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 05:47:49,162 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5104, 5.0577, 5.0324, 5.5152, 5.0838, 4.7156, 5.4825, 4.5465], device='cuda:1'), covar=tensor([0.0310, 0.0852, 0.0301, 0.0360, 0.0747, 0.0431, 0.0438, 0.0490], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0275, 0.0193, 0.0191, 0.0183, 0.0155, 0.0284, 0.0167], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:47:56,012 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.208e+02 2.673e+02 3.606e+02 1.148e+03, threshold=5.346e+02, percent-clipped=3.0 2022-12-08 05:48:19,976 INFO [train.py:873] (1/4) Epoch 14, batch 3900, loss[loss=0.1025, simple_loss=0.1397, pruned_loss=0.03268, over 14581.00 frames. ], tot_loss[loss=0.1153, simple_loss=0.1486, pruned_loss=0.04104, over 1925894.58 frames. ], batch size: 43, lr: 5.64e-03, grad_scale: 4.0 2022-12-08 05:48:44,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9390, 2.5609, 2.7204, 1.7832, 2.4423, 2.6679, 2.9496, 2.4118], device='cuda:1'), covar=tensor([0.0717, 0.0930, 0.0892, 0.1621, 0.1063, 0.0794, 0.0675, 0.1382], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0175, 0.0137, 0.0126, 0.0139, 0.0152, 0.0128, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:49:13,896 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3527, 1.4076, 1.3246, 1.5351, 1.5328, 0.9622, 1.3653, 1.3168], device='cuda:1'), covar=tensor([0.0648, 0.0624, 0.0687, 0.0576, 0.0549, 0.0839, 0.0662, 0.0708], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0029, 0.0033, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:49:22,101 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102276.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:49:26,437 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.029e+02 2.446e+02 2.972e+02 7.280e+02, threshold=4.892e+02, percent-clipped=2.0 2022-12-08 05:49:46,916 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8463, 1.6442, 1.9400, 1.6639, 2.0404, 1.8303, 1.7043, 1.8881], device='cuda:1'), covar=tensor([0.0616, 0.1440, 0.0386, 0.0511, 0.0453, 0.0720, 0.0319, 0.0369], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0314, 0.0394, 0.0301, 0.0372, 0.0323, 0.0364, 0.0300], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:49:49,241 INFO [train.py:873] (1/4) Epoch 14, batch 4000, loss[loss=0.1487, simple_loss=0.1415, pruned_loss=0.07791, over 1237.00 frames. ], tot_loss[loss=0.1147, simple_loss=0.1487, pruned_loss=0.04033, over 1991293.97 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:49:59,027 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9441, 3.0165, 3.0942, 2.9523, 3.0075, 2.8666, 1.4892, 2.8068], device='cuda:1'), covar=tensor([0.0421, 0.0405, 0.0399, 0.0409, 0.0359, 0.0789, 0.2785, 0.0333], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0175, 0.0144, 0.0143, 0.0205, 0.0140, 0.0161, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 05:50:06,321 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 05:50:10,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9318, 3.0027, 3.0838, 2.9520, 3.0232, 2.8808, 1.4694, 2.8007], device='cuda:1'), covar=tensor([0.0449, 0.0465, 0.0445, 0.0462, 0.0386, 0.0745, 0.2909, 0.0384], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0176, 0.0144, 0.0144, 0.0206, 0.0141, 0.0161, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 05:50:18,960 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102340.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:50:21,410 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102343.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:50:25,927 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7697, 3.5564, 4.1144, 3.2132, 2.8240, 3.6002, 1.9841, 3.6128], device='cuda:1'), covar=tensor([0.0926, 0.1083, 0.0522, 0.1866, 0.1991, 0.0885, 0.3706, 0.1062], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0091, 0.0100, 0.0117, 0.0088, 0.0123, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 05:50:55,302 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.024e+02 2.365e+02 2.996e+02 7.946e+02, threshold=4.730e+02, percent-clipped=3.0 2022-12-08 05:51:01,415 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102388.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:51:18,481 INFO [train.py:873] (1/4) Epoch 14, batch 4100, loss[loss=0.1669, simple_loss=0.1583, pruned_loss=0.08774, over 2611.00 frames. ], tot_loss[loss=0.1151, simple_loss=0.1486, pruned_loss=0.04081, over 1984506.53 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:52:23,527 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.525e+01 2.118e+02 2.762e+02 3.343e+02 8.705e+02, threshold=5.523e+02, percent-clipped=3.0 2022-12-08 05:52:46,040 INFO [train.py:873] (1/4) Epoch 14, batch 4200, loss[loss=0.1471, simple_loss=0.1336, pruned_loss=0.08023, over 1248.00 frames. ], tot_loss[loss=0.1151, simple_loss=0.1488, pruned_loss=0.04069, over 1961229.07 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:52:53,033 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9950, 1.9906, 1.9185, 2.0828, 1.9671, 1.2833, 1.8542, 2.0584], device='cuda:1'), covar=tensor([0.0651, 0.0652, 0.0756, 0.0645, 0.0949, 0.0785, 0.0631, 0.0504], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:53:11,144 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-08 05:53:11,518 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2520, 1.7348, 2.4151, 2.0392, 2.2879, 1.6759, 1.9680, 2.2520], device='cuda:1'), covar=tensor([0.1948, 0.3240, 0.0624, 0.2137, 0.1256, 0.2337, 0.1093, 0.0993], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0205, 0.0210, 0.0279, 0.0229, 0.0208, 0.0204, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:53:16,692 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 05:53:36,499 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6047, 1.5970, 1.7659, 1.5212, 1.3940, 1.3608, 1.3778, 1.0910], device='cuda:1'), covar=tensor([0.0173, 0.0199, 0.0141, 0.0192, 0.0193, 0.0308, 0.0244, 0.0368], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0019, 0.0017, 0.0018, 0.0018, 0.0029, 0.0024, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:53:38,579 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 05:53:45,692 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102576.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:53:49,868 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.062e+02 2.234e+02 2.939e+02 3.624e+02 9.310e+02, threshold=5.879e+02, percent-clipped=3.0 2022-12-08 05:54:13,398 INFO [train.py:873] (1/4) Epoch 14, batch 4300, loss[loss=0.1019, simple_loss=0.144, pruned_loss=0.0299, over 14550.00 frames. ], tot_loss[loss=0.1169, simple_loss=0.1499, pruned_loss=0.04191, over 1950537.85 frames. ], batch size: 43, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:54:27,865 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102624.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:54:32,724 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1795, 1.5459, 1.7818, 1.6396, 1.6204, 1.6642, 1.3371, 1.2855], device='cuda:1'), covar=tensor([0.1575, 0.1187, 0.0403, 0.0634, 0.1164, 0.0879, 0.1625, 0.1767], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0086, 0.0068, 0.0072, 0.0097, 0.0084, 0.0099, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 05:54:44,491 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:54:59,223 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 05:55:16,065 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.389e+01 1.915e+02 2.366e+02 2.869e+02 5.820e+02, threshold=4.732e+02, percent-clipped=0.0 2022-12-08 05:55:25,243 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:55:29,379 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8892, 2.7222, 2.7488, 2.9170, 2.8320, 2.8240, 2.9861, 2.5164], device='cuda:1'), covar=tensor([0.0621, 0.1168, 0.0611, 0.0621, 0.0798, 0.0598, 0.0657, 0.0659], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0270, 0.0190, 0.0188, 0.0180, 0.0153, 0.0279, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:55:38,796 INFO [train.py:873] (1/4) Epoch 14, batch 4400, loss[loss=0.09775, simple_loss=0.1452, pruned_loss=0.02513, over 14213.00 frames. ], tot_loss[loss=0.1176, simple_loss=0.1502, pruned_loss=0.04245, over 1977236.90 frames. ], batch size: 32, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:55:41,849 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2022-12-08 05:55:48,460 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8598, 4.5245, 4.3650, 4.8739, 4.5648, 4.2917, 4.8636, 4.1136], device='cuda:1'), covar=tensor([0.0331, 0.0905, 0.0378, 0.0369, 0.0759, 0.0602, 0.0485, 0.0478], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0271, 0.0191, 0.0189, 0.0181, 0.0153, 0.0280, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 05:56:44,117 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.029e+02 2.612e+02 3.153e+02 6.081e+02, threshold=5.224e+02, percent-clipped=1.0 2022-12-08 05:57:07,852 INFO [train.py:873] (1/4) Epoch 14, batch 4500, loss[loss=0.111, simple_loss=0.1498, pruned_loss=0.03613, over 14269.00 frames. ], tot_loss[loss=0.1163, simple_loss=0.1496, pruned_loss=0.04144, over 2013515.13 frames. ], batch size: 46, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:57:20,629 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-08 05:57:23,345 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 05:57:46,911 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102851.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:58:00,727 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5492, 1.4933, 2.7392, 1.4446, 2.7486, 2.6705, 2.0358, 2.8643], device='cuda:1'), covar=tensor([0.0312, 0.2616, 0.0383, 0.1995, 0.0418, 0.0538, 0.1045, 0.0271], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0155, 0.0158, 0.0168, 0.0168, 0.0177, 0.0133, 0.0148], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:58:12,667 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.344e+02 2.809e+02 3.466e+02 7.090e+02, threshold=5.617e+02, percent-clipped=4.0 2022-12-08 05:58:19,045 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102887.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:58:34,047 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1390, 1.1914, 0.9583, 0.9881, 1.1689, 0.6227, 1.0471, 1.0786], device='cuda:1'), covar=tensor([0.0533, 0.0453, 0.0519, 0.0564, 0.0437, 0.0459, 0.0816, 0.0839], device='cuda:1'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 05:58:36,490 INFO [train.py:873] (1/4) Epoch 14, batch 4600, loss[loss=0.1602, simple_loss=0.1484, pruned_loss=0.08598, over 1231.00 frames. ], tot_loss[loss=0.1177, simple_loss=0.1506, pruned_loss=0.04238, over 1975903.75 frames. ], batch size: 100, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:58:40,730 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102912.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:59:12,715 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102948.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:59:27,848 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8525, 1.3395, 1.9913, 1.2931, 1.8796, 2.0063, 1.6507, 2.0709], device='cuda:1'), covar=tensor([0.0391, 0.2257, 0.0593, 0.1988, 0.0718, 0.0649, 0.1173, 0.0439], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0156, 0.0158, 0.0168, 0.0168, 0.0178, 0.0133, 0.0148], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:59:37,565 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9699, 2.7766, 4.9712, 3.4056, 4.7369, 2.2895, 3.6691, 4.7988], device='cuda:1'), covar=tensor([0.0453, 0.3611, 0.0410, 0.5944, 0.0479, 0.3269, 0.1211, 0.0236], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0207, 0.0212, 0.0278, 0.0230, 0.0211, 0.0205, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 05:59:38,371 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3702, 2.2330, 4.9793, 4.5348, 4.3057, 5.0865, 4.8183, 5.1044], device='cuda:1'), covar=tensor([0.1461, 0.1390, 0.0089, 0.0191, 0.0217, 0.0114, 0.0110, 0.0099], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0161, 0.0131, 0.0171, 0.0148, 0.0143, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 05:59:41,554 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 1.979e+02 2.486e+02 3.145e+02 5.919e+02, threshold=4.973e+02, percent-clipped=1.0 2022-12-08 05:59:53,332 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2022-12-08 06:00:04,660 INFO [train.py:873] (1/4) Epoch 14, batch 4700, loss[loss=0.1219, simple_loss=0.1586, pruned_loss=0.04255, over 14207.00 frames. ], tot_loss[loss=0.1172, simple_loss=0.1502, pruned_loss=0.04214, over 1959327.51 frames. ], batch size: 89, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:00:16,409 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5838, 4.6298, 4.8077, 4.1715, 4.6239, 4.9896, 1.9716, 4.3178], device='cuda:1'), covar=tensor([0.0313, 0.0410, 0.0514, 0.0426, 0.0379, 0.0159, 0.3238, 0.0392], device='cuda:1'), in_proj_covar=tensor([0.0167, 0.0172, 0.0143, 0.0140, 0.0203, 0.0138, 0.0157, 0.0191], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 06:01:10,015 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.100e+02 2.622e+02 3.309e+02 6.848e+02, threshold=5.244e+02, percent-clipped=4.0 2022-12-08 06:01:10,159 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6287, 5.0579, 4.8938, 5.5231, 5.1744, 4.7220, 5.5155, 4.5913], device='cuda:1'), covar=tensor([0.0272, 0.1095, 0.0370, 0.0399, 0.0684, 0.0380, 0.0438, 0.0452], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0269, 0.0188, 0.0188, 0.0180, 0.0152, 0.0276, 0.0164], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 06:01:25,857 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103098.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:01:29,283 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2013, 3.0350, 2.4035, 3.3298, 3.1102, 3.2099, 2.8850, 2.4131], device='cuda:1'), covar=tensor([0.0850, 0.1447, 0.3296, 0.0776, 0.1053, 0.1180, 0.1415, 0.3247], device='cuda:1'), in_proj_covar=tensor([0.0274, 0.0293, 0.0262, 0.0274, 0.0319, 0.0300, 0.0257, 0.0247], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:01:33,435 INFO [train.py:873] (1/4) Epoch 14, batch 4800, loss[loss=0.1094, simple_loss=0.1513, pruned_loss=0.03372, over 14538.00 frames. ], tot_loss[loss=0.1158, simple_loss=0.1493, pruned_loss=0.04118, over 2010149.53 frames. ], batch size: 34, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:01:55,885 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:02:02,754 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 06:02:19,461 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103159.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:02:39,048 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.240e+02 2.787e+02 3.398e+02 6.860e+02, threshold=5.574e+02, percent-clipped=4.0 2022-12-08 06:02:49,654 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:02,615 INFO [train.py:873] (1/4) Epoch 14, batch 4900, loss[loss=0.1225, simple_loss=0.1641, pruned_loss=0.04046, over 14289.00 frames. ], tot_loss[loss=0.118, simple_loss=0.1508, pruned_loss=0.04259, over 1998028.86 frames. ], batch size: 25, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:03:02,685 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103207.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:15,748 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7296, 3.4550, 3.1431, 2.3074, 3.2111, 3.4700, 3.7870, 2.9902], device='cuda:1'), covar=tensor([0.0503, 0.0876, 0.0865, 0.1488, 0.0759, 0.0655, 0.0685, 0.1069], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0176, 0.0137, 0.0126, 0.0139, 0.0151, 0.0127, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:1') 2022-12-08 06:03:33,961 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103243.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:48,606 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 06:04:06,508 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.061e+02 2.525e+02 3.223e+02 5.765e+02, threshold=5.049e+02, percent-clipped=2.0 2022-12-08 06:04:08,385 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0522, 2.6316, 3.5709, 2.4018, 2.2263, 3.1505, 1.6761, 3.1957], device='cuda:1'), covar=tensor([0.1173, 0.1544, 0.0624, 0.2266, 0.2141, 0.0804, 0.3136, 0.0676], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0100, 0.0091, 0.0098, 0.0117, 0.0088, 0.0122, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 06:04:29,068 INFO [train.py:873] (1/4) Epoch 14, batch 5000, loss[loss=0.212, simple_loss=0.1809, pruned_loss=0.1215, over 1195.00 frames. ], tot_loss[loss=0.1179, simple_loss=0.15, pruned_loss=0.04291, over 1879521.48 frames. ], batch size: 100, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:05:00,386 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 06:05:20,829 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103366.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:05:33,574 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.256e+02 2.699e+02 3.304e+02 5.196e+02, threshold=5.398e+02, percent-clipped=1.0 2022-12-08 06:05:40,707 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 06:05:55,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7785, 1.4047, 3.6973, 1.8590, 3.7044, 3.8861, 2.7064, 4.1405], device='cuda:1'), covar=tensor([0.0238, 0.3160, 0.0402, 0.2142, 0.0618, 0.0342, 0.0902, 0.0162], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0155, 0.0159, 0.0169, 0.0168, 0.0179, 0.0133, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:05:56,321 INFO [train.py:873] (1/4) Epoch 14, batch 5100, loss[loss=0.1139, simple_loss=0.1491, pruned_loss=0.03936, over 14259.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1497, pruned_loss=0.04178, over 1986695.42 frames. ], batch size: 76, lr: 5.60e-03, grad_scale: 4.0 2022-12-08 06:06:13,819 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103427.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 06:06:37,604 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:06:52,205 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8249, 2.0466, 2.1567, 1.8684, 1.9088, 1.7417, 1.6427, 1.2870], device='cuda:1'), covar=tensor([0.0264, 0.0292, 0.0233, 0.0250, 0.0321, 0.0260, 0.0269, 0.0498], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0017, 0.0018, 0.0018, 0.0030, 0.0024, 0.0029], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 06:07:01,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.364e+02 2.994e+02 3.604e+02 8.056e+02, threshold=5.989e+02, percent-clipped=5.0 2022-12-08 06:07:06,911 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103488.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:07:23,312 INFO [train.py:873] (1/4) Epoch 14, batch 5200, loss[loss=0.1653, simple_loss=0.1751, pruned_loss=0.07775, over 8597.00 frames. ], tot_loss[loss=0.1169, simple_loss=0.15, pruned_loss=0.04192, over 1962712.25 frames. ], batch size: 100, lr: 5.60e-03, grad_scale: 8.0 2022-12-08 06:07:23,433 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103507.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:07:23,466 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3033, 1.9233, 2.2581, 1.5932, 1.8804, 2.2551, 2.1946, 2.0141], device='cuda:1'), covar=tensor([0.0884, 0.0816, 0.0909, 0.1612, 0.1465, 0.0938, 0.0865, 0.1484], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0176, 0.0138, 0.0127, 0.0142, 0.0153, 0.0129, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:07:30,753 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2022-12-08 06:07:55,027 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103543.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:04,929 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:28,605 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 2.069e+02 2.548e+02 2.976e+02 7.652e+02, threshold=5.095e+02, percent-clipped=1.0 2022-12-08 06:08:36,193 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:50,465 INFO [train.py:873] (1/4) Epoch 14, batch 5300, loss[loss=0.09697, simple_loss=0.1287, pruned_loss=0.0326, over 13912.00 frames. ], tot_loss[loss=0.1158, simple_loss=0.1492, pruned_loss=0.04122, over 1980435.49 frames. ], batch size: 20, lr: 5.60e-03, grad_scale: 8.0 2022-12-08 06:09:37,574 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3728, 3.2952, 3.1640, 3.4229, 3.0293, 2.9271, 3.4166, 3.3040], device='cuda:1'), covar=tensor([0.0730, 0.0904, 0.1013, 0.0696, 0.1102, 0.0821, 0.0742, 0.0823], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0138, 0.0144, 0.0158, 0.0146, 0.0123, 0.0166, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 06:09:55,897 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.008e+02 2.559e+02 3.052e+02 6.405e+02, threshold=5.118e+02, percent-clipped=1.0 2022-12-08 06:10:15,064 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.63 vs. limit=5.0 2022-12-08 06:10:17,828 INFO [train.py:873] (1/4) Epoch 14, batch 5400, loss[loss=0.08777, simple_loss=0.1306, pruned_loss=0.02246, over 14046.00 frames. ], tot_loss[loss=0.1165, simple_loss=0.1496, pruned_loss=0.04173, over 1976462.71 frames. ], batch size: 26, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:10:30,731 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103722.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:10:47,280 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:10:49,214 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2022-12-08 06:10:59,263 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103754.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:09,117 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:24,939 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 06:11:25,001 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.268e+02 2.622e+02 3.465e+02 5.931e+02, threshold=5.245e+02, percent-clipped=4.0 2022-12-08 06:11:29,621 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:41,163 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103801.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:41,942 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103802.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:46,663 INFO [train.py:873] (1/4) Epoch 14, batch 5500, loss[loss=0.1833, simple_loss=0.191, pruned_loss=0.08778, over 10347.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.149, pruned_loss=0.04193, over 1939443.13 frames. ], batch size: 100, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:12:03,754 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103826.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:12,177 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103836.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:47,302 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103876.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:53,011 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.148e+02 2.635e+02 3.186e+02 7.478e+02, threshold=5.269e+02, percent-clipped=1.0 2022-12-08 06:13:14,657 INFO [train.py:873] (1/4) Epoch 14, batch 5600, loss[loss=0.09516, simple_loss=0.1391, pruned_loss=0.02562, over 14307.00 frames. ], tot_loss[loss=0.1173, simple_loss=0.1498, pruned_loss=0.0424, over 1966157.46 frames. ], batch size: 39, lr: 5.59e-03, grad_scale: 8.0 2022-12-08 06:13:18,288 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5950, 4.0474, 3.7278, 3.5411, 2.8293, 3.9554, 3.6944, 2.4315], device='cuda:1'), covar=tensor([0.1418, 0.0479, 0.1185, 0.1253, 0.0899, 0.0405, 0.0908, 0.1797], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0085, 0.0067, 0.0071, 0.0097, 0.0083, 0.0099, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:13:41,212 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103937.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:14:21,890 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.071e+02 2.747e+02 3.245e+02 6.809e+02, threshold=5.494e+02, percent-clipped=2.0 2022-12-08 06:14:27,082 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103990.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:14:28,792 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8010, 2.4595, 2.9606, 2.0047, 1.9856, 2.6310, 1.4432, 2.6356], device='cuda:1'), covar=tensor([0.0821, 0.1454, 0.0739, 0.1981, 0.2327, 0.1089, 0.3918, 0.1042], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0099, 0.0090, 0.0099, 0.0115, 0.0086, 0.0120, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 06:14:33,057 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103997.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:14:42,174 INFO [train.py:873] (1/4) Epoch 14, batch 5700, loss[loss=0.1694, simple_loss=0.1553, pruned_loss=0.09172, over 1350.00 frames. ], tot_loss[loss=0.1167, simple_loss=0.1497, pruned_loss=0.04187, over 1952083.40 frames. ], batch size: 100, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:14:55,484 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104022.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:15:18,678 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=104049.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:20,379 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:26,935 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:29,183 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2331, 4.3669, 4.5093, 4.0943, 4.4270, 4.6870, 1.7731, 4.0746], device='cuda:1'), covar=tensor([0.0400, 0.0403, 0.0499, 0.0509, 0.0380, 0.0219, 0.3651, 0.0372], device='cuda:1'), in_proj_covar=tensor([0.0169, 0.0173, 0.0144, 0.0142, 0.0203, 0.0139, 0.0159, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 06:15:36,679 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104070.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:15:49,004 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.112e+02 2.664e+02 3.363e+02 7.605e+02, threshold=5.327e+02, percent-clipped=4.0 2022-12-08 06:15:55,142 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7230, 1.9162, 2.6316, 2.1294, 2.5675, 2.5722, 2.4555, 2.3096], device='cuda:1'), covar=tensor([0.0719, 0.2815, 0.0860, 0.1749, 0.0691, 0.1123, 0.0933, 0.1432], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0314, 0.0395, 0.0303, 0.0378, 0.0324, 0.0363, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:15:56,259 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 06:15:59,755 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104096.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:08,845 INFO [train.py:873] (1/4) Epoch 14, batch 5800, loss[loss=0.1437, simple_loss=0.1635, pruned_loss=0.06195, over 11144.00 frames. ], tot_loss[loss=0.1174, simple_loss=0.1503, pruned_loss=0.04226, over 2009251.18 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 4.0 2022-12-08 06:16:11,526 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:20,701 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104121.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:21,650 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1558, 3.7241, 3.0582, 4.4194, 4.1611, 4.1951, 3.6030, 2.9985], device='cuda:1'), covar=tensor([0.0829, 0.1294, 0.3243, 0.0585, 0.0963, 0.1597, 0.1217, 0.3429], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0292, 0.0265, 0.0276, 0.0321, 0.0299, 0.0257, 0.0247], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:16:52,660 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4638, 5.2035, 4.9505, 5.4787, 4.9921, 4.8473, 5.4972, 5.2603], device='cuda:1'), covar=tensor([0.0550, 0.0617, 0.0681, 0.0448, 0.0815, 0.0431, 0.0457, 0.0616], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0138, 0.0143, 0.0158, 0.0144, 0.0122, 0.0167, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 06:17:16,434 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.209e+02 2.706e+02 3.376e+02 1.574e+03, threshold=5.411e+02, percent-clipped=5.0 2022-12-08 06:17:36,927 INFO [train.py:873] (1/4) Epoch 14, batch 5900, loss[loss=0.1244, simple_loss=0.1562, pruned_loss=0.04632, over 14469.00 frames. ], tot_loss[loss=0.117, simple_loss=0.1496, pruned_loss=0.04219, over 1953058.77 frames. ], batch size: 49, lr: 5.58e-03, grad_scale: 4.0 2022-12-08 06:17:43,874 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1525, 2.0885, 1.8237, 1.9119, 2.0700, 2.1052, 2.1007, 2.1094], device='cuda:1'), covar=tensor([0.1073, 0.0977, 0.2641, 0.2455, 0.1383, 0.1267, 0.1655, 0.1116], device='cuda:1'), in_proj_covar=tensor([0.0385, 0.0269, 0.0449, 0.0567, 0.0345, 0.0440, 0.0391, 0.0384], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:17:50,280 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5851, 2.8239, 4.4057, 3.2836, 4.2676, 4.2715, 4.1084, 3.8079], device='cuda:1'), covar=tensor([0.0841, 0.3026, 0.0866, 0.1923, 0.0838, 0.0885, 0.1826, 0.1800], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0313, 0.0392, 0.0300, 0.0375, 0.0320, 0.0360, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:17:58,448 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104232.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 06:18:00,285 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8041, 3.2816, 2.9358, 3.1955, 2.4710, 3.2292, 3.0462, 1.6692], device='cuda:1'), covar=tensor([0.1628, 0.0715, 0.1344, 0.0731, 0.0960, 0.0552, 0.0950, 0.2241], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0085, 0.0067, 0.0072, 0.0097, 0.0084, 0.0099, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:18:13,401 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7788, 1.3636, 1.6956, 1.2829, 1.4864, 1.8008, 1.5427, 1.5602], device='cuda:1'), covar=tensor([0.0578, 0.0638, 0.0551, 0.0739, 0.1255, 0.0700, 0.0697, 0.1337], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0176, 0.0138, 0.0127, 0.0140, 0.0152, 0.0129, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:18:43,106 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.217e+02 2.661e+02 3.155e+02 4.728e+02, threshold=5.321e+02, percent-clipped=0.0 2022-12-08 06:18:53,865 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8065, 1.3383, 1.6632, 1.2640, 1.4982, 1.8355, 1.4947, 1.5413], device='cuda:1'), covar=tensor([0.0666, 0.0812, 0.0582, 0.0819, 0.1362, 0.0777, 0.0732, 0.1639], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0176, 0.0137, 0.0127, 0.0140, 0.0152, 0.0128, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:19:00,197 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 06:19:04,079 INFO [train.py:873] (1/4) Epoch 14, batch 6000, loss[loss=0.1052, simple_loss=0.1313, pruned_loss=0.03948, over 4991.00 frames. ], tot_loss[loss=0.1167, simple_loss=0.1495, pruned_loss=0.04198, over 1968153.06 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 8.0 2022-12-08 06:19:04,079 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 06:19:08,172 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7544, 3.6790, 3.3311, 2.6846, 3.3018, 3.6796, 3.9358, 3.1732], device='cuda:1'), covar=tensor([0.0602, 0.1034, 0.0960, 0.1251, 0.0720, 0.0610, 0.0663, 0.1131], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0176, 0.0137, 0.0126, 0.0140, 0.0151, 0.0128, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:19:12,428 INFO [train.py:905] (1/4) Epoch 14, validation: loss=0.1346, simple_loss=0.1729, pruned_loss=0.04809, over 857387.00 frames. 2022-12-08 06:19:12,428 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 06:19:17,071 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9841, 3.2718, 4.7216, 3.6155, 4.6267, 4.5709, 4.4322, 4.0830], device='cuda:1'), covar=tensor([0.0463, 0.2510, 0.0636, 0.1601, 0.0701, 0.0758, 0.1421, 0.1569], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0315, 0.0394, 0.0303, 0.0377, 0.0323, 0.0363, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:19:39,465 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4370, 1.7147, 2.6547, 2.1624, 2.5628, 1.7907, 2.1223, 2.4749], device='cuda:1'), covar=tensor([0.1662, 0.3088, 0.0548, 0.2353, 0.0901, 0.2344, 0.0985, 0.0766], device='cuda:1'), in_proj_covar=tensor([0.0249, 0.0204, 0.0208, 0.0273, 0.0226, 0.0208, 0.0203, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:19:45,973 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:19:52,278 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:02,102 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7799, 0.8658, 0.7131, 0.8145, 0.8322, 0.3452, 0.7124, 0.8775], device='cuda:1'), covar=tensor([0.0421, 0.0471, 0.0510, 0.0443, 0.0311, 0.0302, 0.1084, 0.0678], device='cuda:1'), in_proj_covar=tensor([0.0031, 0.0031, 0.0033, 0.0029, 0.0030, 0.0043, 0.0031, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 06:20:18,637 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.150e+02 2.718e+02 3.385e+02 8.730e+02, threshold=5.435e+02, percent-clipped=4.0 2022-12-08 06:20:29,283 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:37,598 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104405.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:39,255 INFO [train.py:873] (1/4) Epoch 14, batch 6100, loss[loss=0.1496, simple_loss=0.1528, pruned_loss=0.07316, over 3867.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.1491, pruned_loss=0.04165, over 1960399.93 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 8.0 2022-12-08 06:20:51,282 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104421.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:06,750 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2022-12-08 06:21:11,697 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104444.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:28,544 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5535, 5.2802, 4.9552, 5.1421, 5.0794, 5.4373, 5.5173, 5.4506], device='cuda:1'), covar=tensor([0.0753, 0.0373, 0.1908, 0.2463, 0.0693, 0.0744, 0.0762, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0387, 0.0269, 0.0454, 0.0574, 0.0349, 0.0446, 0.0395, 0.0387], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:21:32,836 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104469.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:38,620 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7839, 1.7174, 2.9577, 2.1729, 2.8206, 1.7537, 2.3276, 2.7524], device='cuda:1'), covar=tensor([0.1176, 0.4024, 0.0679, 0.3804, 0.0887, 0.3239, 0.1269, 0.0690], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0204, 0.0211, 0.0274, 0.0228, 0.0210, 0.0204, 0.0213], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:21:46,358 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.353e+02 3.023e+02 3.584e+02 7.529e+02, threshold=6.047e+02, percent-clipped=6.0 2022-12-08 06:22:05,691 INFO [train.py:873] (1/4) Epoch 14, batch 6200, loss[loss=0.09644, simple_loss=0.1393, pruned_loss=0.0268, over 14410.00 frames. ], tot_loss[loss=0.1169, simple_loss=0.1496, pruned_loss=0.04213, over 1933304.21 frames. ], batch size: 31, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:22:27,814 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104532.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 06:22:55,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1269, 2.1851, 2.9055, 2.3795, 2.8046, 2.8815, 2.7092, 2.5299], device='cuda:1'), covar=tensor([0.0857, 0.2422, 0.0887, 0.1691, 0.0676, 0.0941, 0.1006, 0.1531], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0314, 0.0394, 0.0302, 0.0374, 0.0323, 0.0362, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:23:09,238 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104580.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:23:12,641 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.145e+02 2.527e+02 3.084e+02 9.249e+02, threshold=5.054e+02, percent-clipped=2.0 2022-12-08 06:23:33,265 INFO [train.py:873] (1/4) Epoch 14, batch 6300, loss[loss=0.168, simple_loss=0.1824, pruned_loss=0.07682, over 8638.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1495, pruned_loss=0.0419, over 1913056.70 frames. ], batch size: 100, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:24:07,397 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:13,568 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:40,912 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.535e+01 2.052e+02 2.556e+02 3.229e+02 6.200e+02, threshold=5.113e+02, percent-clipped=1.0 2022-12-08 06:24:45,024 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 06:24:49,226 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104694.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:55,245 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104701.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:58,749 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104705.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:25:00,315 INFO [train.py:873] (1/4) Epoch 14, batch 6400, loss[loss=0.1544, simple_loss=0.1461, pruned_loss=0.08136, over 1330.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1495, pruned_loss=0.0418, over 1889145.49 frames. ], batch size: 100, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:25:27,927 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9394, 5.3363, 5.3277, 5.8481, 5.4229, 4.8287, 5.8307, 4.8003], device='cuda:1'), covar=tensor([0.0300, 0.0878, 0.0358, 0.0409, 0.0704, 0.0360, 0.0453, 0.0535], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0270, 0.0194, 0.0191, 0.0184, 0.0154, 0.0283, 0.0166], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 06:25:40,657 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104753.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:26:07,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.032e+02 2.617e+02 3.308e+02 7.760e+02, threshold=5.234e+02, percent-clipped=3.0 2022-12-08 06:26:27,854 INFO [train.py:873] (1/4) Epoch 14, batch 6500, loss[loss=0.1184, simple_loss=0.1523, pruned_loss=0.04229, over 14576.00 frames. ], tot_loss[loss=0.1163, simple_loss=0.1491, pruned_loss=0.04172, over 1895820.45 frames. ], batch size: 49, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:26:53,749 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7958, 1.3948, 3.7384, 1.5873, 3.7329, 3.8992, 2.8564, 4.1642], device='cuda:1'), covar=tensor([0.0271, 0.3568, 0.0496, 0.2640, 0.0542, 0.0412, 0.0839, 0.0227], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0157, 0.0161, 0.0170, 0.0169, 0.0179, 0.0135, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:27:35,547 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.183e+02 2.623e+02 3.627e+02 5.274e+02, threshold=5.247e+02, percent-clipped=1.0 2022-12-08 06:27:54,697 INFO [train.py:873] (1/4) Epoch 14, batch 6600, loss[loss=0.1142, simple_loss=0.1363, pruned_loss=0.04602, over 3852.00 frames. ], tot_loss[loss=0.1158, simple_loss=0.1489, pruned_loss=0.04132, over 1911441.02 frames. ], batch size: 100, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:28:14,269 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 06:28:15,266 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 06:29:02,029 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.257e+02 2.787e+02 3.469e+02 5.148e+02, threshold=5.574e+02, percent-clipped=0.0 2022-12-08 06:29:27,170 INFO [train.py:873] (1/4) Epoch 14, batch 6700, loss[loss=0.1021, simple_loss=0.1425, pruned_loss=0.03081, over 14275.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.1496, pruned_loss=0.04139, over 1967809.47 frames. ], batch size: 44, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:29:30,772 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9404, 1.7422, 4.1559, 3.9791, 3.9127, 4.2429, 3.4947, 4.2546], device='cuda:1'), covar=tensor([0.1536, 0.1576, 0.0133, 0.0205, 0.0227, 0.0128, 0.0291, 0.0121], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0130, 0.0169, 0.0146, 0.0141, 0.0123, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 06:29:52,917 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-08 06:29:53,800 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-08 06:30:21,256 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7902, 1.5969, 2.0723, 1.6203, 1.9628, 1.4719, 1.6501, 1.9274], device='cuda:1'), covar=tensor([0.2828, 0.2110, 0.0581, 0.1588, 0.1131, 0.1461, 0.1262, 0.0646], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0204, 0.0212, 0.0275, 0.0229, 0.0210, 0.0204, 0.0211], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:30:34,208 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7580, 1.4410, 2.5171, 2.2568, 2.3745, 2.5079, 1.7001, 2.5101], device='cuda:1'), covar=tensor([0.0994, 0.1245, 0.0205, 0.0466, 0.0484, 0.0214, 0.0714, 0.0251], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0129, 0.0168, 0.0145, 0.0140, 0.0123, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 06:30:35,607 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.155e+02 2.089e+02 2.481e+02 3.177e+02 6.321e+02, threshold=4.962e+02, percent-clipped=1.0 2022-12-08 06:30:54,206 INFO [train.py:873] (1/4) Epoch 14, batch 6800, loss[loss=0.1205, simple_loss=0.1565, pruned_loss=0.04221, over 14593.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.1493, pruned_loss=0.04154, over 1977083.18 frames. ], batch size: 23, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:31:13,461 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8023, 2.3722, 3.6305, 2.7296, 3.6143, 3.4441, 3.4467, 3.0665], device='cuda:1'), covar=tensor([0.0806, 0.2904, 0.0959, 0.1865, 0.0849, 0.1011, 0.1494, 0.1715], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0314, 0.0396, 0.0303, 0.0375, 0.0324, 0.0363, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:31:31,096 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2022-12-08 06:31:50,616 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0857, 2.2368, 2.3297, 2.4250, 2.0345, 2.3115, 2.2580, 1.3985], device='cuda:1'), covar=tensor([0.1157, 0.0891, 0.0715, 0.0526, 0.0981, 0.0732, 0.1130, 0.2112], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0086, 0.0068, 0.0071, 0.0098, 0.0084, 0.0100, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:31:54,146 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 06:32:01,691 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.161e+02 2.579e+02 3.505e+02 6.694e+02, threshold=5.159e+02, percent-clipped=6.0 2022-12-08 06:32:21,682 INFO [train.py:873] (1/4) Epoch 14, batch 6900, loss[loss=0.1106, simple_loss=0.1516, pruned_loss=0.03476, over 14003.00 frames. ], tot_loss[loss=0.1172, simple_loss=0.15, pruned_loss=0.04224, over 1993082.51 frames. ], batch size: 22, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:32:33,121 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 06:32:53,655 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7071, 2.3709, 4.7214, 3.2047, 4.4387, 2.1490, 3.6034, 4.5266], device='cuda:1'), covar=tensor([0.0459, 0.3803, 0.0336, 0.5514, 0.0630, 0.3217, 0.1084, 0.0350], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0205, 0.0211, 0.0277, 0.0227, 0.0209, 0.0205, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:33:29,079 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.377e+02 2.807e+02 3.459e+02 9.593e+02, threshold=5.615e+02, percent-clipped=4.0 2022-12-08 06:33:48,170 INFO [train.py:873] (1/4) Epoch 14, batch 7000, loss[loss=0.07668, simple_loss=0.1163, pruned_loss=0.01853, over 13584.00 frames. ], tot_loss[loss=0.1172, simple_loss=0.1498, pruned_loss=0.04224, over 1938617.30 frames. ], batch size: 17, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:34:45,719 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 06:34:56,768 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.288e+01 2.265e+02 2.868e+02 3.572e+02 1.076e+03, threshold=5.736e+02, percent-clipped=4.0 2022-12-08 06:34:59,578 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105388.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:35:16,755 INFO [train.py:873] (1/4) Epoch 14, batch 7100, loss[loss=0.1753, simple_loss=0.1566, pruned_loss=0.09702, over 1185.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.1495, pruned_loss=0.04144, over 2008831.84 frames. ], batch size: 100, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:35:22,187 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2660, 5.1157, 4.7439, 5.2648, 4.9227, 4.7586, 5.2927, 5.1179], device='cuda:1'), covar=tensor([0.0505, 0.0640, 0.0766, 0.0535, 0.0688, 0.0466, 0.0585, 0.0603], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0136, 0.0140, 0.0155, 0.0141, 0.0119, 0.0162, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 06:35:23,065 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4206, 3.2133, 3.0182, 3.1287, 3.3138, 3.3247, 3.4075, 3.3668], device='cuda:1'), covar=tensor([0.0979, 0.0727, 0.2443, 0.2811, 0.0993, 0.1064, 0.1189, 0.1012], device='cuda:1'), in_proj_covar=tensor([0.0378, 0.0262, 0.0438, 0.0556, 0.0338, 0.0436, 0.0382, 0.0378], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:35:53,733 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105449.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:35:55,066 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 06:36:25,709 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.137e+02 2.673e+02 3.373e+02 4.675e+02, threshold=5.347e+02, percent-clipped=1.0 2022-12-08 06:36:25,953 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:36:39,544 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105502.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:36:43,629 INFO [train.py:873] (1/4) Epoch 14, batch 7200, loss[loss=0.09732, simple_loss=0.1375, pruned_loss=0.02858, over 14634.00 frames. ], tot_loss[loss=0.1161, simple_loss=0.1491, pruned_loss=0.04157, over 1924576.32 frames. ], batch size: 22, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:36:44,415 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1664, 1.3108, 1.4089, 1.0161, 0.9133, 1.1772, 1.0096, 1.2620], device='cuda:1'), covar=tensor([0.1622, 0.2611, 0.0918, 0.2321, 0.2710, 0.1081, 0.2084, 0.1141], device='cuda:1'), in_proj_covar=tensor([0.0083, 0.0098, 0.0089, 0.0098, 0.0115, 0.0086, 0.0120, 0.0090], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 06:36:58,288 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3795, 1.3029, 2.5385, 1.5190, 2.4417, 2.4796, 1.9527, 2.6344], device='cuda:1'), covar=tensor([0.0291, 0.2416, 0.0399, 0.1689, 0.0505, 0.0523, 0.1049, 0.0284], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0154, 0.0158, 0.0167, 0.0166, 0.0174, 0.0132, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:37:14,522 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.9948, 1.1044, 0.9898, 1.1142, 1.1526, 0.7851, 0.9781, 1.0694], device='cuda:1'), covar=tensor([0.0854, 0.0809, 0.0801, 0.0536, 0.0463, 0.0834, 0.1047, 0.0867], device='cuda:1'), in_proj_covar=tensor([0.0032, 0.0031, 0.0034, 0.0029, 0.0031, 0.0044, 0.0032, 0.0035], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 06:37:17,874 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105547.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:37:32,146 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105563.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:37:51,600 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.276e+02 2.781e+02 3.572e+02 7.841e+02, threshold=5.562e+02, percent-clipped=2.0 2022-12-08 06:38:10,274 INFO [train.py:873] (1/4) Epoch 14, batch 7300, loss[loss=0.1107, simple_loss=0.1468, pruned_loss=0.03729, over 14296.00 frames. ], tot_loss[loss=0.1157, simple_loss=0.1487, pruned_loss=0.04137, over 1941227.97 frames. ], batch size: 69, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:38:54,071 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0391, 1.9413, 4.5858, 4.2120, 4.1559, 4.6015, 4.0998, 4.6788], device='cuda:1'), covar=tensor([0.1501, 0.1432, 0.0108, 0.0220, 0.0208, 0.0131, 0.0231, 0.0110], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0131, 0.0171, 0.0148, 0.0143, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 06:39:20,008 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.190e+02 2.577e+02 3.146e+02 6.129e+02, threshold=5.153e+02, percent-clipped=1.0 2022-12-08 06:39:35,475 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-08 06:39:38,185 INFO [train.py:873] (1/4) Epoch 14, batch 7400, loss[loss=0.09581, simple_loss=0.1396, pruned_loss=0.02601, over 14269.00 frames. ], tot_loss[loss=0.1151, simple_loss=0.1485, pruned_loss=0.04083, over 2013353.14 frames. ], batch size: 28, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:40:11,138 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105744.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:40:48,154 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.057e+02 2.114e+02 2.694e+02 3.325e+02 8.535e+02, threshold=5.387e+02, percent-clipped=3.0 2022-12-08 06:40:53,354 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105792.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:40:59,950 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7802, 4.5671, 4.2291, 4.4387, 4.5420, 4.7277, 4.8101, 4.8106], device='cuda:1'), covar=tensor([0.1081, 0.0464, 0.2387, 0.2879, 0.0744, 0.0899, 0.1108, 0.0962], device='cuda:1'), in_proj_covar=tensor([0.0380, 0.0262, 0.0443, 0.0562, 0.0339, 0.0437, 0.0384, 0.0382], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:41:00,029 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105799.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:07,345 INFO [train.py:873] (1/4) Epoch 14, batch 7500, loss[loss=0.09153, simple_loss=0.1347, pruned_loss=0.02416, over 14154.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.148, pruned_loss=0.03976, over 2103335.46 frames. ], batch size: 35, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:41:32,435 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 06:41:36,975 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105842.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:44,731 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105853.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:47,024 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105858.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:48,041 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105860.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:42:35,199 INFO [train.py:873] (1/4) Epoch 15, batch 0, loss[loss=0.1192, simple_loss=0.1558, pruned_loss=0.04132, over 14248.00 frames. ], tot_loss[loss=0.1192, simple_loss=0.1558, pruned_loss=0.04132, over 14248.00 frames. ], batch size: 31, lr: 5.35e-03, grad_scale: 8.0 2022-12-08 06:42:35,199 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 06:42:39,181 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9266, 2.6390, 2.8705, 2.0308, 2.4412, 2.7912, 2.9664, 2.4845], device='cuda:1'), covar=tensor([0.0754, 0.0486, 0.0771, 0.1168, 0.1161, 0.0562, 0.0613, 0.1312], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0175, 0.0137, 0.0125, 0.0139, 0.0151, 0.0128, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:42:42,693 INFO [train.py:905] (1/4) Epoch 15, validation: loss=0.1381, simple_loss=0.1782, pruned_loss=0.049, over 857387.00 frames. 2022-12-08 06:42:42,694 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 06:42:58,147 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.674e+01 2.091e+02 3.002e+02 4.364e+02 1.099e+03, threshold=6.004e+02, percent-clipped=13.0 2022-12-08 06:43:21,500 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-08 06:44:07,757 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 06:44:11,357 INFO [train.py:873] (1/4) Epoch 15, batch 100, loss[loss=0.1164, simple_loss=0.1544, pruned_loss=0.03918, over 14430.00 frames. ], tot_loss[loss=0.1163, simple_loss=0.1496, pruned_loss=0.04149, over 879860.57 frames. ], batch size: 53, lr: 5.35e-03, grad_scale: 8.0 2022-12-08 06:44:26,575 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.570e+02 2.979e+02 3.765e+02 9.423e+02, threshold=5.958e+02, percent-clipped=4.0 2022-12-08 06:44:49,996 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9520, 2.2802, 2.3006, 2.3302, 1.9831, 2.3761, 2.2256, 1.2526], device='cuda:1'), covar=tensor([0.1373, 0.1070, 0.0861, 0.0781, 0.0995, 0.0658, 0.1142, 0.2548], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0085, 0.0067, 0.0070, 0.0095, 0.0083, 0.0097, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:45:18,382 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106044.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:45:22,895 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2022-12-08 06:45:37,866 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1659, 1.1525, 1.3124, 1.0050, 1.1417, 0.8906, 0.8291, 0.9128], device='cuda:1'), covar=tensor([0.0214, 0.0285, 0.0193, 0.0292, 0.0288, 0.0516, 0.0315, 0.0545], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 06:45:40,385 INFO [train.py:873] (1/4) Epoch 15, batch 200, loss[loss=0.1184, simple_loss=0.1496, pruned_loss=0.04365, over 6988.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.1492, pruned_loss=0.04159, over 1301984.79 frames. ], batch size: 100, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:45:47,436 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106077.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:45:55,351 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.218e+02 2.817e+02 3.547e+02 5.569e+02, threshold=5.634e+02, percent-clipped=0.0 2022-12-08 06:45:58,286 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1427, 3.8569, 3.4697, 2.6495, 3.5427, 3.8259, 4.2817, 3.4436], device='cuda:1'), covar=tensor([0.0510, 0.1032, 0.0770, 0.1215, 0.0685, 0.0562, 0.0551, 0.0913], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0174, 0.0136, 0.0125, 0.0138, 0.0151, 0.0128, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:46:00,763 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:41,873 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106138.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:45,005 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:48,437 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8944, 3.7015, 3.4366, 3.5725, 3.8189, 3.8095, 3.8961, 3.8787], device='cuda:1'), covar=tensor([0.0912, 0.0598, 0.2132, 0.2586, 0.0803, 0.0928, 0.1022, 0.0873], device='cuda:1'), in_proj_covar=tensor([0.0377, 0.0259, 0.0442, 0.0559, 0.0338, 0.0436, 0.0384, 0.0380], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:46:50,481 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106148.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:56,476 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106155.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:59,189 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:47:08,665 INFO [train.py:873] (1/4) Epoch 15, batch 300, loss[loss=0.1487, simple_loss=0.1422, pruned_loss=0.07767, over 1305.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1479, pruned_loss=0.04001, over 1573399.05 frames. ], batch size: 100, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:47:14,135 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2449, 2.9260, 2.8764, 1.9367, 2.7077, 2.9126, 3.2996, 2.5481], device='cuda:1'), covar=tensor([0.0583, 0.0722, 0.0836, 0.1389, 0.0934, 0.0646, 0.0562, 0.1235], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0176, 0.0138, 0.0126, 0.0141, 0.0152, 0.0129, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:47:23,718 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.130e+02 2.529e+02 3.253e+02 6.005e+02, threshold=5.059e+02, percent-clipped=1.0 2022-12-08 06:47:27,436 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106190.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:47:36,636 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8431, 1.6501, 1.9395, 1.6809, 1.9556, 1.7682, 1.7032, 1.8414], device='cuda:1'), covar=tensor([0.0504, 0.1288, 0.0436, 0.0466, 0.0521, 0.0686, 0.0308, 0.0358], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0317, 0.0394, 0.0300, 0.0373, 0.0324, 0.0361, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:47:42,119 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:48:37,000 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:48:37,735 INFO [train.py:873] (1/4) Epoch 15, batch 400, loss[loss=0.09873, simple_loss=0.1294, pruned_loss=0.03405, over 6001.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1479, pruned_loss=0.04032, over 1688969.62 frames. ], batch size: 100, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:48:52,653 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.076e+02 2.571e+02 3.282e+02 8.496e+02, threshold=5.142e+02, percent-clipped=2.0 2022-12-08 06:49:16,573 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.10 vs. limit=5.0 2022-12-08 06:49:30,211 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:49:36,218 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:50:06,075 INFO [train.py:873] (1/4) Epoch 15, batch 500, loss[loss=0.1683, simple_loss=0.1561, pruned_loss=0.09031, over 1253.00 frames. ], tot_loss[loss=0.1157, simple_loss=0.1488, pruned_loss=0.04132, over 1759177.63 frames. ], batch size: 100, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:50:13,578 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-08 06:50:21,314 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.064e+02 2.651e+02 3.398e+02 5.648e+02, threshold=5.301e+02, percent-clipped=3.0 2022-12-08 06:50:31,310 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106397.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:03,974 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106433.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:17,621 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106448.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:22,866 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:23,657 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106455.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:26,850 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 06:51:36,094 INFO [train.py:873] (1/4) Epoch 15, batch 600, loss[loss=0.0965, simple_loss=0.1395, pruned_loss=0.02673, over 14332.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1479, pruned_loss=0.04004, over 1866314.71 frames. ], batch size: 73, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:51:37,092 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5704, 4.2076, 4.1506, 4.6020, 4.2891, 4.1205, 4.6095, 3.8115], device='cuda:1'), covar=tensor([0.0399, 0.0952, 0.0415, 0.0380, 0.0777, 0.0766, 0.0470, 0.0567], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0267, 0.0191, 0.0188, 0.0180, 0.0150, 0.0277, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 06:51:51,214 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.156e+02 2.802e+02 3.401e+02 5.738e+02, threshold=5.604e+02, percent-clipped=2.0 2022-12-08 06:52:00,112 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106496.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:52:06,415 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106503.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:52:17,470 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106515.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:52:21,741 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8401, 1.6926, 1.9196, 1.7199, 1.9984, 1.7930, 1.6865, 1.8810], device='cuda:1'), covar=tensor([0.0646, 0.1254, 0.0396, 0.0504, 0.0530, 0.0704, 0.0343, 0.0372], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0320, 0.0398, 0.0303, 0.0378, 0.0327, 0.0364, 0.0305], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:52:58,800 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0997, 2.0641, 3.0607, 3.1515, 3.0309, 2.1821, 3.0400, 2.4537], device='cuda:1'), covar=tensor([0.0386, 0.1081, 0.0677, 0.0473, 0.0479, 0.1378, 0.0410, 0.0863], device='cuda:1'), in_proj_covar=tensor([0.0289, 0.0256, 0.0370, 0.0325, 0.0267, 0.0300, 0.0305, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 06:53:05,159 INFO [train.py:873] (1/4) Epoch 15, batch 700, loss[loss=0.1413, simple_loss=0.134, pruned_loss=0.07425, over 1230.00 frames. ], tot_loss[loss=0.1144, simple_loss=0.1479, pruned_loss=0.04041, over 1894250.42 frames. ], batch size: 100, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:53:08,229 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2022-12-08 06:53:19,980 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.992e+01 2.009e+02 2.731e+02 3.273e+02 5.771e+02, threshold=5.463e+02, percent-clipped=3.0 2022-12-08 06:53:53,124 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:54:32,758 INFO [train.py:873] (1/4) Epoch 15, batch 800, loss[loss=0.1379, simple_loss=0.167, pruned_loss=0.05438, over 13988.00 frames. ], tot_loss[loss=0.1153, simple_loss=0.1486, pruned_loss=0.041, over 1945463.32 frames. ], batch size: 22, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:54:47,970 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.230e+02 2.748e+02 3.448e+02 5.498e+02, threshold=5.496e+02, percent-clipped=1.0 2022-12-08 06:54:53,528 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:54:55,246 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6081, 4.0339, 3.1764, 4.9256, 4.3513, 4.6465, 4.1584, 3.4703], device='cuda:1'), covar=tensor([0.0721, 0.1111, 0.3218, 0.0430, 0.1025, 0.1530, 0.0899, 0.2812], device='cuda:1'), in_proj_covar=tensor([0.0273, 0.0290, 0.0262, 0.0275, 0.0318, 0.0295, 0.0254, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:55:04,984 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9332, 1.8547, 2.0531, 1.8743, 1.8326, 1.6971, 1.2653, 1.2384], device='cuda:1'), covar=tensor([0.0252, 0.0453, 0.0272, 0.0297, 0.0333, 0.0364, 0.0357, 0.0593], device='cuda:1'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0019, 0.0030, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 06:55:12,300 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-08 06:55:13,226 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 06:55:24,021 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 06:55:30,057 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106733.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:56:02,202 INFO [train.py:873] (1/4) Epoch 15, batch 900, loss[loss=0.1134, simple_loss=0.1504, pruned_loss=0.03824, over 14182.00 frames. ], tot_loss[loss=0.1148, simple_loss=0.1485, pruned_loss=0.04059, over 1992521.99 frames. ], batch size: 89, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:56:12,631 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:56:16,790 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.159e+02 2.563e+02 3.204e+02 7.042e+02, threshold=5.126e+02, percent-clipped=4.0 2022-12-08 06:56:38,474 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106810.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:57:30,127 INFO [train.py:873] (1/4) Epoch 15, batch 1000, loss[loss=0.1167, simple_loss=0.15, pruned_loss=0.04171, over 11171.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.1487, pruned_loss=0.04084, over 1979570.10 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 06:57:44,838 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.224e+02 2.651e+02 3.456e+02 5.513e+02, threshold=5.301e+02, percent-clipped=2.0 2022-12-08 06:57:56,778 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9091, 1.9146, 1.6744, 1.9398, 1.7582, 1.7838, 1.7983, 1.6971], device='cuda:1'), covar=tensor([0.0871, 0.0825, 0.1813, 0.0646, 0.1095, 0.0633, 0.1474, 0.1124], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0291, 0.0263, 0.0277, 0.0320, 0.0299, 0.0256, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 06:58:18,754 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:58:18,791 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5724, 3.2585, 3.2081, 2.3906, 3.0276, 3.3455, 3.5562, 2.7407], device='cuda:1'), covar=tensor([0.0535, 0.1073, 0.0846, 0.1257, 0.0723, 0.0615, 0.0630, 0.1252], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0177, 0.0139, 0.0128, 0.0142, 0.0154, 0.0131, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 06:58:58,495 INFO [train.py:873] (1/4) Epoch 15, batch 1100, loss[loss=0.1429, simple_loss=0.1672, pruned_loss=0.05933, over 7765.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.1487, pruned_loss=0.04126, over 1922427.87 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 06:59:01,484 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:59:13,541 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.263e+02 2.855e+02 3.604e+02 7.949e+02, threshold=5.710e+02, percent-clipped=5.0 2022-12-08 06:59:19,105 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:59:24,584 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.87 vs. limit=5.0 2022-12-08 06:59:40,220 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6212, 1.6111, 1.6063, 1.6961, 1.7393, 1.0599, 1.5507, 1.5806], device='cuda:1'), covar=tensor([0.0697, 0.0759, 0.0602, 0.0985, 0.0802, 0.0913, 0.0712, 0.0694], device='cuda:1'), in_proj_covar=tensor([0.0031, 0.0031, 0.0035, 0.0029, 0.0031, 0.0044, 0.0032, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:00:02,186 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:00:03,165 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8711, 1.5489, 3.3762, 3.0949, 3.1842, 3.3693, 2.6330, 3.3231], device='cuda:1'), covar=tensor([0.1529, 0.1590, 0.0158, 0.0314, 0.0308, 0.0186, 0.0423, 0.0197], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0131, 0.0170, 0.0149, 0.0143, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 07:00:19,179 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0050, 4.1042, 4.3322, 3.7600, 4.1982, 4.3018, 1.5965, 3.9571], device='cuda:1'), covar=tensor([0.0332, 0.0387, 0.0368, 0.0504, 0.0277, 0.0293, 0.3173, 0.0274], device='cuda:1'), in_proj_covar=tensor([0.0167, 0.0172, 0.0144, 0.0144, 0.0203, 0.0139, 0.0157, 0.0190], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 07:00:28,131 INFO [train.py:873] (1/4) Epoch 15, batch 1200, loss[loss=0.1432, simple_loss=0.1445, pruned_loss=0.0709, over 2627.00 frames. ], tot_loss[loss=0.1157, simple_loss=0.1487, pruned_loss=0.04137, over 1884918.55 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 07:00:42,788 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 2.082e+02 2.691e+02 3.251e+02 9.387e+02, threshold=5.382e+02, percent-clipped=1.0 2022-12-08 07:01:01,036 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 07:01:04,249 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:30,672 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107140.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:01:38,630 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1717, 3.9745, 3.8841, 4.2516, 3.7784, 3.4895, 4.2389, 4.0882], device='cuda:1'), covar=tensor([0.0677, 0.0868, 0.0833, 0.0603, 0.0853, 0.0712, 0.0593, 0.0683], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0140, 0.0144, 0.0158, 0.0146, 0.0123, 0.0166, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:01:41,302 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107152.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:46,381 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:56,715 INFO [train.py:873] (1/4) Epoch 15, batch 1300, loss[loss=0.1371, simple_loss=0.1492, pruned_loss=0.06252, over 3918.00 frames. ], tot_loss[loss=0.1155, simple_loss=0.1484, pruned_loss=0.0413, over 1898926.85 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 07:02:03,631 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.59 vs. limit=5.0 2022-12-08 07:02:12,107 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.125e+02 2.037e+02 2.513e+02 3.265e+02 6.018e+02, threshold=5.026e+02, percent-clipped=1.0 2022-12-08 07:02:15,816 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0322, 1.1719, 1.0171, 1.0728, 1.2165, 0.7187, 0.9416, 1.1070], device='cuda:1'), covar=tensor([0.0639, 0.0610, 0.0913, 0.0550, 0.0497, 0.0755, 0.1347, 0.0715], device='cuda:1'), in_proj_covar=tensor([0.0031, 0.0031, 0.0035, 0.0029, 0.0031, 0.0044, 0.0032, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:02:25,622 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107201.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:02:36,981 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107213.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:03:00,901 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7180, 1.9709, 2.0984, 2.1319, 1.9316, 2.0881, 1.8777, 1.3130], device='cuda:1'), covar=tensor([0.0941, 0.1065, 0.0627, 0.0666, 0.1112, 0.0763, 0.1181, 0.2265], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0087, 0.0068, 0.0071, 0.0098, 0.0085, 0.0101, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:03:04,498 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9358, 2.8471, 2.1476, 3.0038, 2.8315, 2.8607, 2.6288, 2.2550], device='cuda:1'), covar=tensor([0.0887, 0.1461, 0.2768, 0.0957, 0.1082, 0.1167, 0.1260, 0.2730], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0291, 0.0263, 0.0275, 0.0320, 0.0299, 0.0257, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:03:18,944 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9412, 3.5755, 2.7858, 4.1875, 4.0448, 3.9971, 3.5625, 3.0665], device='cuda:1'), covar=tensor([0.1037, 0.1228, 0.3295, 0.0533, 0.0725, 0.1384, 0.1098, 0.2737], device='cuda:1'), in_proj_covar=tensor([0.0276, 0.0293, 0.0264, 0.0276, 0.0321, 0.0301, 0.0258, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:03:26,074 INFO [train.py:873] (1/4) Epoch 15, batch 1400, loss[loss=0.1529, simple_loss=0.1733, pruned_loss=0.06619, over 14262.00 frames. ], tot_loss[loss=0.115, simple_loss=0.1483, pruned_loss=0.0408, over 1956176.70 frames. ], batch size: 60, lr: 5.31e-03, grad_scale: 4.0 2022-12-08 07:03:41,696 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 2.191e+02 2.804e+02 3.719e+02 1.162e+03, threshold=5.608e+02, percent-clipped=5.0 2022-12-08 07:04:03,608 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0253, 2.1580, 3.0136, 3.1305, 3.0260, 2.2217, 3.0504, 2.4225], device='cuda:1'), covar=tensor([0.0444, 0.1011, 0.0678, 0.0453, 0.0482, 0.1339, 0.0404, 0.0837], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0258, 0.0375, 0.0329, 0.0269, 0.0303, 0.0306, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:04:54,154 INFO [train.py:873] (1/4) Epoch 15, batch 1500, loss[loss=0.1291, simple_loss=0.142, pruned_loss=0.05813, over 2582.00 frames. ], tot_loss[loss=0.1149, simple_loss=0.1481, pruned_loss=0.04085, over 1958961.67 frames. ], batch size: 100, lr: 5.31e-03, grad_scale: 4.0 2022-12-08 07:05:10,370 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.250e+02 2.599e+02 3.364e+02 1.423e+03, threshold=5.197e+02, percent-clipped=1.0 2022-12-08 07:05:21,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7559, 3.0305, 2.9480, 3.0725, 2.4443, 3.1201, 2.9212, 1.6091], device='cuda:1'), covar=tensor([0.1456, 0.0907, 0.1051, 0.0517, 0.0982, 0.0432, 0.0947, 0.2353], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0087, 0.0068, 0.0071, 0.0098, 0.0085, 0.0100, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:05:49,905 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0151, 3.6072, 2.8232, 4.2956, 4.0393, 4.0867, 3.5571, 2.9764], device='cuda:1'), covar=tensor([0.0742, 0.1187, 0.3077, 0.0430, 0.0867, 0.1325, 0.1111, 0.2825], device='cuda:1'), in_proj_covar=tensor([0.0276, 0.0291, 0.0262, 0.0274, 0.0320, 0.0299, 0.0257, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:06:17,743 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 07:06:23,610 INFO [train.py:873] (1/4) Epoch 15, batch 1600, loss[loss=0.1371, simple_loss=0.1531, pruned_loss=0.06055, over 7804.00 frames. ], tot_loss[loss=0.1141, simple_loss=0.1477, pruned_loss=0.04022, over 1963749.38 frames. ], batch size: 100, lr: 5.31e-03, grad_scale: 8.0 2022-12-08 07:06:39,091 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.559e+01 2.229e+02 2.676e+02 3.216e+02 6.513e+02, threshold=5.353e+02, percent-clipped=3.0 2022-12-08 07:06:47,612 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107496.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:06:58,132 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107508.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:06:59,886 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:07:51,201 INFO [train.py:873] (1/4) Epoch 15, batch 1700, loss[loss=0.1255, simple_loss=0.1302, pruned_loss=0.06043, over 2607.00 frames. ], tot_loss[loss=0.1144, simple_loss=0.1482, pruned_loss=0.04029, over 1993231.14 frames. ], batch size: 100, lr: 5.31e-03, grad_scale: 8.0 2022-12-08 07:07:53,384 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107571.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:08:07,305 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 2.065e+02 2.454e+02 2.883e+02 5.170e+02, threshold=4.908e+02, percent-clipped=0.0 2022-12-08 07:08:09,225 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4209, 1.0961, 1.2502, 0.8546, 1.1217, 1.4131, 1.0447, 1.1022], device='cuda:1'), covar=tensor([0.0466, 0.0887, 0.0772, 0.0468, 0.0931, 0.0727, 0.0554, 0.1252], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0177, 0.0139, 0.0128, 0.0142, 0.0154, 0.0131, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:08:46,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2413, 3.0442, 2.8188, 2.9211, 3.1790, 3.1632, 3.2024, 3.2086], device='cuda:1'), covar=tensor([0.0921, 0.0642, 0.2042, 0.2492, 0.0872, 0.1117, 0.1112, 0.0828], device='cuda:1'), in_proj_covar=tensor([0.0385, 0.0263, 0.0447, 0.0566, 0.0343, 0.0439, 0.0392, 0.0384], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:09:11,894 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:09:20,339 INFO [train.py:873] (1/4) Epoch 15, batch 1800, loss[loss=0.1145, simple_loss=0.1498, pruned_loss=0.0396, over 14293.00 frames. ], tot_loss[loss=0.1141, simple_loss=0.148, pruned_loss=0.0401, over 1971470.12 frames. ], batch size: 39, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:09:35,244 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.283e+02 3.065e+02 3.850e+02 1.119e+03, threshold=6.129e+02, percent-clipped=8.0 2022-12-08 07:09:55,454 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6609, 2.4275, 4.7016, 3.2210, 4.4165, 2.2379, 3.5247, 4.4956], device='cuda:1'), covar=tensor([0.0550, 0.3943, 0.0377, 0.5603, 0.0635, 0.3218, 0.1186, 0.0335], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0205, 0.0212, 0.0274, 0.0226, 0.0207, 0.0204, 0.0212], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:09:55,986 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 07:10:04,778 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:10:12,762 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6172, 2.3489, 3.3480, 2.5992, 3.4971, 3.3454, 3.2552, 2.9108], device='cuda:1'), covar=tensor([0.0920, 0.2743, 0.1095, 0.1926, 0.0865, 0.0940, 0.1238, 0.1639], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0315, 0.0394, 0.0302, 0.0377, 0.0323, 0.0363, 0.0301], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:10:20,187 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107737.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:10:48,099 INFO [train.py:873] (1/4) Epoch 15, batch 1900, loss[loss=0.1389, simple_loss=0.1663, pruned_loss=0.05575, over 14139.00 frames. ], tot_loss[loss=0.1161, simple_loss=0.149, pruned_loss=0.04159, over 1968701.10 frames. ], batch size: 99, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:11:04,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.201e+02 2.753e+02 3.141e+02 7.986e+02, threshold=5.507e+02, percent-clipped=3.0 2022-12-08 07:11:12,173 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107796.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:11:13,922 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107798.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:11:22,811 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107808.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:11:29,485 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1221, 3.2085, 3.4116, 3.2020, 3.3055, 2.8837, 1.4478, 3.1326], device='cuda:1'), covar=tensor([0.0401, 0.0431, 0.0367, 0.0399, 0.0345, 0.0733, 0.3002, 0.0305], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0171, 0.0143, 0.0143, 0.0201, 0.0139, 0.0157, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 07:11:54,591 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107844.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:12:05,295 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107856.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:12:09,970 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.53 vs. limit=5.0 2022-12-08 07:12:13,831 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107866.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:12:16,483 INFO [train.py:873] (1/4) Epoch 15, batch 2000, loss[loss=0.1216, simple_loss=0.1424, pruned_loss=0.05037, over 3866.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.149, pruned_loss=0.04108, over 2025172.64 frames. ], batch size: 100, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:12:19,008 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2022-12-08 07:12:31,885 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 1.995e+02 2.775e+02 3.641e+02 9.267e+02, threshold=5.550e+02, percent-clipped=5.0 2022-12-08 07:13:33,471 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2774, 2.5280, 2.6538, 2.6496, 2.2262, 2.5568, 2.4041, 1.4296], device='cuda:1'), covar=tensor([0.1080, 0.0849, 0.0554, 0.0442, 0.0941, 0.0548, 0.1082, 0.2074], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0087, 0.0068, 0.0070, 0.0098, 0.0085, 0.0100, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:13:42,198 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7591, 2.1957, 3.8170, 3.9188, 3.7739, 2.2542, 3.8907, 2.7658], device='cuda:1'), covar=tensor([0.0520, 0.1251, 0.0861, 0.0584, 0.0502, 0.1835, 0.0515, 0.1184], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0256, 0.0373, 0.0328, 0.0269, 0.0302, 0.0307, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:13:43,634 INFO [train.py:873] (1/4) Epoch 15, batch 2100, loss[loss=0.1216, simple_loss=0.1515, pruned_loss=0.04587, over 10349.00 frames. ], tot_loss[loss=0.1148, simple_loss=0.1484, pruned_loss=0.0406, over 2017553.74 frames. ], batch size: 100, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:13:58,706 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0407, 1.1266, 0.8900, 1.1163, 1.1632, 0.7148, 0.9214, 1.0556], device='cuda:1'), covar=tensor([0.0682, 0.0595, 0.0797, 0.0571, 0.0446, 0.0796, 0.1004, 0.0668], device='cuda:1'), in_proj_covar=tensor([0.0031, 0.0031, 0.0034, 0.0029, 0.0031, 0.0044, 0.0032, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:13:59,441 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 1.955e+02 2.546e+02 3.190e+02 6.620e+02, threshold=5.091e+02, percent-clipped=1.0 2022-12-08 07:14:06,718 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5854, 2.2932, 2.9761, 1.8448, 2.0326, 2.6868, 1.6154, 2.4924], device='cuda:1'), covar=tensor([0.1425, 0.1658, 0.0716, 0.2877, 0.2388, 0.0906, 0.3571, 0.1269], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0092, 0.0098, 0.0117, 0.0089, 0.0121, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 07:14:11,725 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5709, 2.6095, 2.7605, 2.7446, 2.6569, 2.4579, 1.4682, 2.4427], device='cuda:1'), covar=tensor([0.0525, 0.0503, 0.0458, 0.0383, 0.0429, 0.1039, 0.2803, 0.0394], device='cuda:1'), in_proj_covar=tensor([0.0166, 0.0171, 0.0143, 0.0142, 0.0201, 0.0138, 0.0157, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 07:14:18,590 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-08 07:14:24,593 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:14:33,898 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7314, 1.9058, 2.0513, 1.7374, 2.0370, 1.6085, 1.5472, 1.1392], device='cuda:1'), covar=tensor([0.0332, 0.0504, 0.0227, 0.0423, 0.0236, 0.0324, 0.0324, 0.0581], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:14:48,829 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.07 vs. limit=5.0 2022-12-08 07:15:11,289 INFO [train.py:873] (1/4) Epoch 15, batch 2200, loss[loss=0.1293, simple_loss=0.1397, pruned_loss=0.05945, over 3898.00 frames. ], tot_loss[loss=0.116, simple_loss=0.1488, pruned_loss=0.04161, over 1963355.54 frames. ], batch size: 100, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:15:26,510 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.145e+02 2.568e+02 3.392e+02 6.463e+02, threshold=5.135e+02, percent-clipped=2.0 2022-12-08 07:15:31,535 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1099, 3.1749, 3.3333, 3.1629, 3.1983, 2.7581, 1.4431, 3.0141], device='cuda:1'), covar=tensor([0.0391, 0.0425, 0.0384, 0.0403, 0.0334, 0.1044, 0.3041, 0.0307], device='cuda:1'), in_proj_covar=tensor([0.0165, 0.0169, 0.0142, 0.0141, 0.0200, 0.0137, 0.0156, 0.0187], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 07:15:32,314 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108093.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:16:36,136 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108166.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:16:39,019 INFO [train.py:873] (1/4) Epoch 15, batch 2300, loss[loss=0.09385, simple_loss=0.1401, pruned_loss=0.02382, over 14287.00 frames. ], tot_loss[loss=0.1145, simple_loss=0.1478, pruned_loss=0.04055, over 1941783.05 frames. ], batch size: 44, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:16:53,591 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.53 vs. limit=5.0 2022-12-08 07:16:54,803 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.030e+02 2.596e+02 3.385e+02 5.799e+02, threshold=5.191e+02, percent-clipped=3.0 2022-12-08 07:16:55,878 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6139, 2.5691, 2.0352, 2.6483, 2.4898, 2.5592, 2.3760, 2.0953], device='cuda:1'), covar=tensor([0.1211, 0.1125, 0.2741, 0.0875, 0.1210, 0.0785, 0.1494, 0.1916], device='cuda:1'), in_proj_covar=tensor([0.0274, 0.0287, 0.0261, 0.0276, 0.0319, 0.0296, 0.0254, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:17:18,545 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108214.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:17:44,732 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8029, 2.8196, 2.7019, 2.9009, 2.4528, 2.5614, 2.9092, 2.7996], device='cuda:1'), covar=tensor([0.0737, 0.0978, 0.0812, 0.0746, 0.1210, 0.0802, 0.0708, 0.0769], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0137, 0.0141, 0.0155, 0.0142, 0.0120, 0.0162, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:18:06,534 INFO [train.py:873] (1/4) Epoch 15, batch 2400, loss[loss=0.1223, simple_loss=0.1434, pruned_loss=0.05053, over 5960.00 frames. ], tot_loss[loss=0.115, simple_loss=0.148, pruned_loss=0.04096, over 1892227.33 frames. ], batch size: 100, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:18:08,133 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-08 07:18:17,774 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 07:18:22,283 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.197e+02 2.869e+02 3.470e+02 6.788e+02, threshold=5.738e+02, percent-clipped=3.0 2022-12-08 07:18:47,568 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7300, 4.5047, 4.3333, 4.7251, 4.3434, 4.0198, 4.7846, 4.6191], device='cuda:1'), covar=tensor([0.0596, 0.0713, 0.0703, 0.0569, 0.0691, 0.0553, 0.0481, 0.0634], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0136, 0.0141, 0.0154, 0.0142, 0.0120, 0.0163, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:18:47,657 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:18:48,815 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2022-12-08 07:18:55,842 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108325.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:29,553 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:34,871 INFO [train.py:873] (1/4) Epoch 15, batch 2500, loss[loss=0.1028, simple_loss=0.1436, pruned_loss=0.03107, over 14420.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1471, pruned_loss=0.0399, over 1948115.06 frames. ], batch size: 73, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:19:35,469 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 07:19:50,276 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108386.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:50,976 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.088e+02 2.601e+02 3.226e+02 6.971e+02, threshold=5.203e+02, percent-clipped=3.0 2022-12-08 07:19:56,290 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108393.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:20:25,411 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9655, 3.8498, 3.7090, 4.1006, 3.8738, 3.5855, 4.1520, 3.4492], device='cuda:1'), covar=tensor([0.0652, 0.0875, 0.0434, 0.0448, 0.0704, 0.1304, 0.0510, 0.0551], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0273, 0.0196, 0.0195, 0.0183, 0.0153, 0.0283, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 07:20:31,730 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:20:38,936 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:21:03,650 INFO [train.py:873] (1/4) Epoch 15, batch 2600, loss[loss=0.1274, simple_loss=0.1383, pruned_loss=0.05827, over 3839.00 frames. ], tot_loss[loss=0.1146, simple_loss=0.1473, pruned_loss=0.04089, over 1928424.15 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:21:19,264 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.775e+01 2.028e+02 2.663e+02 3.413e+02 6.606e+02, threshold=5.327e+02, percent-clipped=6.0 2022-12-08 07:21:24,984 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108493.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:22:08,762 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5768, 1.7114, 4.3262, 2.1076, 4.3127, 4.6857, 4.0783, 5.0011], device='cuda:1'), covar=tensor([0.0214, 0.3083, 0.0434, 0.2285, 0.0306, 0.0340, 0.0358, 0.0151], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0157, 0.0161, 0.0170, 0.0167, 0.0177, 0.0134, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:22:16,772 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5693, 3.0023, 4.3347, 3.3629, 4.3443, 4.3146, 4.1403, 3.7996], device='cuda:1'), covar=tensor([0.0746, 0.2639, 0.0980, 0.1585, 0.0870, 0.0842, 0.1418, 0.1522], device='cuda:1'), in_proj_covar=tensor([0.0357, 0.0317, 0.0397, 0.0301, 0.0377, 0.0324, 0.0366, 0.0305], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:22:21,492 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 07:22:25,309 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0003, 1.5155, 4.1802, 1.8511, 3.9843, 4.3557, 3.7985, 4.5870], device='cuda:1'), covar=tensor([0.0371, 0.4273, 0.0523, 0.3099, 0.0542, 0.0366, 0.0484, 0.0309], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0160, 0.0169, 0.0167, 0.0177, 0.0133, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:22:32,111 INFO [train.py:873] (1/4) Epoch 15, batch 2700, loss[loss=0.1614, simple_loss=0.145, pruned_loss=0.08892, over 1277.00 frames. ], tot_loss[loss=0.1145, simple_loss=0.1481, pruned_loss=0.04044, over 1964717.84 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 4.0 2022-12-08 07:22:39,965 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:22:49,625 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.154e+02 2.629e+02 3.316e+02 6.561e+02, threshold=5.258e+02, percent-clipped=1.0 2022-12-08 07:23:33,965 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:24:01,225 INFO [train.py:873] (1/4) Epoch 15, batch 2800, loss[loss=0.1897, simple_loss=0.1681, pruned_loss=0.1057, over 1205.00 frames. ], tot_loss[loss=0.1144, simple_loss=0.148, pruned_loss=0.04036, over 1974568.25 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:24:12,082 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108681.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:24:17,714 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.202e+02 2.137e+02 2.731e+02 3.357e+02 8.128e+02, threshold=5.462e+02, percent-clipped=3.0 2022-12-08 07:24:30,024 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 07:24:50,764 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3178, 3.1760, 2.4870, 3.4210, 3.3457, 3.3883, 3.0408, 2.5308], device='cuda:1'), covar=tensor([0.0962, 0.1300, 0.3225, 0.0843, 0.0902, 0.1276, 0.1235, 0.2868], device='cuda:1'), in_proj_covar=tensor([0.0275, 0.0288, 0.0261, 0.0277, 0.0320, 0.0298, 0.0254, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:1') 2022-12-08 07:25:05,548 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 07:25:06,509 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 07:25:29,355 INFO [train.py:873] (1/4) Epoch 15, batch 2900, loss[loss=0.09498, simple_loss=0.1392, pruned_loss=0.02537, over 14575.00 frames. ], tot_loss[loss=0.115, simple_loss=0.1485, pruned_loss=0.04071, over 2003851.07 frames. ], batch size: 22, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:25:45,502 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.119e+02 2.529e+02 3.118e+02 5.786e+02, threshold=5.058e+02, percent-clipped=1.0 2022-12-08 07:25:45,656 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108788.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:25:48,281 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108791.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:26:11,421 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.14 vs. limit=5.0 2022-12-08 07:26:41,489 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2512, 2.6052, 4.0704, 3.0871, 4.0678, 3.9756, 3.8266, 3.4771], device='cuda:1'), covar=tensor([0.0819, 0.3047, 0.0948, 0.1751, 0.0833, 0.0944, 0.1549, 0.1683], device='cuda:1'), in_proj_covar=tensor([0.0358, 0.0320, 0.0397, 0.0302, 0.0379, 0.0326, 0.0366, 0.0306], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:26:42,252 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:26:56,661 INFO [train.py:873] (1/4) Epoch 15, batch 3000, loss[loss=0.09661, simple_loss=0.1428, pruned_loss=0.02522, over 14483.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.1485, pruned_loss=0.0409, over 1918417.60 frames. ], batch size: 49, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:26:56,661 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 07:27:00,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5635, 2.7653, 4.5108, 3.4738, 4.5896, 4.4705, 4.2069, 3.9091], device='cuda:1'), covar=tensor([0.0678, 0.3265, 0.0744, 0.1711, 0.0593, 0.0716, 0.1174, 0.1682], device='cuda:1'), in_proj_covar=tensor([0.0358, 0.0320, 0.0397, 0.0302, 0.0379, 0.0325, 0.0366, 0.0306], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:27:03,522 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8400, 2.6404, 2.7523, 2.8021, 2.7000, 2.6495, 2.8373, 2.7012], device='cuda:1'), covar=tensor([0.0313, 0.0852, 0.0481, 0.0524, 0.0666, 0.0433, 0.0637, 0.0438], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0268, 0.0193, 0.0192, 0.0180, 0.0150, 0.0278, 0.0165], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 07:27:05,082 INFO [train.py:905] (1/4) Epoch 15, validation: loss=0.1368, simple_loss=0.1737, pruned_loss=0.04998, over 857387.00 frames. 2022-12-08 07:27:05,083 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 07:27:21,818 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 2.224e+02 2.739e+02 3.295e+02 6.855e+02, threshold=5.478e+02, percent-clipped=4.0 2022-12-08 07:27:36,144 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 07:28:00,933 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:28:32,782 INFO [train.py:873] (1/4) Epoch 15, batch 3100, loss[loss=0.1525, simple_loss=0.1518, pruned_loss=0.07664, over 2643.00 frames. ], tot_loss[loss=0.1148, simple_loss=0.1481, pruned_loss=0.04076, over 1926348.07 frames. ], batch size: 100, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:28:39,232 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 07:28:42,893 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:28:48,816 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.020e+02 2.542e+02 3.088e+02 8.012e+02, threshold=5.083e+02, percent-clipped=4.0 2022-12-08 07:29:25,232 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109029.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:29:32,498 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109037.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:30:00,244 INFO [train.py:873] (1/4) Epoch 15, batch 3200, loss[loss=0.106, simple_loss=0.1199, pruned_loss=0.04606, over 2721.00 frames. ], tot_loss[loss=0.115, simple_loss=0.1484, pruned_loss=0.04074, over 1916228.87 frames. ], batch size: 100, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:30:05,421 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 07:30:16,994 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.174e+02 2.531e+02 2.915e+02 6.335e+02, threshold=5.062e+02, percent-clipped=1.0 2022-12-08 07:30:17,239 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:30:25,653 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109098.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:30:58,797 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:31:08,409 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:31:27,857 INFO [train.py:873] (1/4) Epoch 15, batch 3300, loss[loss=0.1416, simple_loss=0.1307, pruned_loss=0.07626, over 1299.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.1477, pruned_loss=0.03989, over 2002123.80 frames. ], batch size: 100, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:31:40,696 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 07:31:44,097 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.026e+02 2.178e+02 2.677e+02 3.330e+02 7.070e+02, threshold=5.354e+02, percent-clipped=2.0 2022-12-08 07:31:49,539 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109194.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:00,612 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-08 07:32:23,629 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:43,245 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109255.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:55,126 INFO [train.py:873] (1/4) Epoch 15, batch 3400, loss[loss=0.1413, simple_loss=0.1315, pruned_loss=0.07556, over 1300.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1477, pruned_loss=0.0397, over 2055772.24 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 4.0 2022-12-08 07:33:05,763 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:33:12,975 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.873e+01 2.037e+02 2.446e+02 3.255e+02 7.608e+02, threshold=4.893e+02, percent-clipped=4.0 2022-12-08 07:33:13,501 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 07:33:17,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109294.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:33:30,414 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 07:33:38,174 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:34:03,804 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2906, 1.6197, 4.1754, 1.9786, 4.1131, 4.4108, 3.6010, 4.7331], device='cuda:1'), covar=tensor([0.0236, 0.3113, 0.0373, 0.2118, 0.0380, 0.0333, 0.0538, 0.0152], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0154, 0.0157, 0.0166, 0.0166, 0.0175, 0.0131, 0.0148], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:34:10,852 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109355.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:34:16,410 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 07:34:23,254 INFO [train.py:873] (1/4) Epoch 15, batch 3500, loss[loss=0.1351, simple_loss=0.1631, pruned_loss=0.05358, over 8609.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.1475, pruned_loss=0.0399, over 1972770.78 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 4.0 2022-12-08 07:34:32,373 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109379.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:34:40,312 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.264e+02 2.745e+02 3.644e+02 1.157e+03, threshold=5.490e+02, percent-clipped=6.0 2022-12-08 07:34:44,038 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109393.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:35:31,952 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109447.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:35:32,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-08 07:35:38,996 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2711, 1.5685, 1.6912, 1.7747, 1.6611, 1.7121, 1.3747, 1.2262], device='cuda:1'), covar=tensor([0.1124, 0.1428, 0.0605, 0.0774, 0.1114, 0.0996, 0.2082, 0.2316], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0088, 0.0069, 0.0072, 0.0098, 0.0086, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:35:50,855 INFO [train.py:873] (1/4) Epoch 15, batch 3600, loss[loss=0.1053, simple_loss=0.1466, pruned_loss=0.03196, over 14409.00 frames. ], tot_loss[loss=0.1136, simple_loss=0.148, pruned_loss=0.03961, over 2010689.48 frames. ], batch size: 41, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:36:08,169 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8064, 1.6131, 2.0618, 1.6845, 1.8846, 1.4989, 1.6572, 1.8965], device='cuda:1'), covar=tensor([0.2919, 0.1838, 0.0648, 0.1435, 0.1419, 0.1273, 0.1099, 0.0777], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0206, 0.0215, 0.0279, 0.0232, 0.0208, 0.0207, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:36:08,874 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.104e+02 2.724e+02 3.497e+02 9.894e+02, threshold=5.448e+02, percent-clipped=5.0 2022-12-08 07:36:14,323 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:37:02,967 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109550.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:37:08,677 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2022-12-08 07:37:19,586 INFO [train.py:873] (1/4) Epoch 15, batch 3700, loss[loss=0.1055, simple_loss=0.1434, pruned_loss=0.0338, over 14145.00 frames. ], tot_loss[loss=0.1146, simple_loss=0.1486, pruned_loss=0.04027, over 2045847.59 frames. ], batch size: 84, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:37:20,881 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4106, 5.1676, 4.8792, 5.4506, 4.9102, 4.5801, 5.4861, 5.2008], device='cuda:1'), covar=tensor([0.0598, 0.0738, 0.0792, 0.0556, 0.0799, 0.0560, 0.0514, 0.0666], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0142, 0.0146, 0.0160, 0.0147, 0.0124, 0.0169, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:37:28,847 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109579.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:37:37,109 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.068e+02 2.547e+02 3.153e+02 5.371e+02, threshold=5.093e+02, percent-clipped=0.0 2022-12-08 07:38:22,994 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109640.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:32,141 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:39,862 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:48,151 INFO [train.py:873] (1/4) Epoch 15, batch 3800, loss[loss=0.1044, simple_loss=0.1413, pruned_loss=0.03375, over 13544.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.1479, pruned_loss=0.03983, over 2009644.44 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:38:53,033 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109674.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:38:53,908 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0872, 4.5264, 4.8085, 5.1182, 4.8051, 4.1991, 5.1251, 4.0574], device='cuda:1'), covar=tensor([0.0880, 0.1745, 0.0780, 0.0846, 0.1072, 0.0823, 0.0875, 0.1062], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0270, 0.0197, 0.0194, 0.0183, 0.0155, 0.0285, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 07:39:06,076 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.193e+02 2.777e+02 3.474e+02 7.448e+02, threshold=5.555e+02, percent-clipped=3.0 2022-12-08 07:39:10,140 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109693.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:39:33,979 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:39:52,263 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109741.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:39:56,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8519, 1.9903, 2.7623, 2.2258, 2.6777, 2.6451, 2.4457, 2.3475], device='cuda:1'), covar=tensor([0.0941, 0.2910, 0.0973, 0.1676, 0.0716, 0.1109, 0.0924, 0.1547], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0315, 0.0392, 0.0298, 0.0373, 0.0322, 0.0359, 0.0301], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:40:16,359 INFO [train.py:873] (1/4) Epoch 15, batch 3900, loss[loss=0.1612, simple_loss=0.167, pruned_loss=0.07774, over 3899.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1469, pruned_loss=0.03996, over 1888491.49 frames. ], batch size: 100, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:40:34,143 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.235e+02 2.763e+02 3.267e+02 1.043e+03, threshold=5.525e+02, percent-clipped=3.0 2022-12-08 07:41:28,434 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:41:30,191 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1408, 1.8740, 2.0351, 1.9213, 1.9093, 1.8677, 1.6231, 1.5884], device='cuda:1'), covar=tensor([0.0281, 0.0972, 0.0449, 0.0562, 0.0308, 0.0447, 0.0372, 0.0485], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0032, 0.0026, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:41:45,243 INFO [train.py:873] (1/4) Epoch 15, batch 4000, loss[loss=0.1062, simple_loss=0.1411, pruned_loss=0.03564, over 12755.00 frames. ], tot_loss[loss=0.1136, simple_loss=0.1468, pruned_loss=0.04015, over 1833164.51 frames. ], batch size: 100, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:42:02,600 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 2.144e+02 2.556e+02 3.390e+02 6.366e+02, threshold=5.112e+02, percent-clipped=4.0 2022-12-08 07:42:10,899 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109898.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:42:18,562 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8788, 1.4748, 1.8349, 1.2795, 1.5698, 1.9101, 1.6623, 1.6724], device='cuda:1'), covar=tensor([0.0787, 0.0793, 0.0936, 0.1261, 0.1525, 0.1009, 0.0941, 0.1522], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0172, 0.0138, 0.0127, 0.0140, 0.0152, 0.0130, 0.0139], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 07:42:43,621 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109935.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:42:57,035 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:13,346 INFO [train.py:873] (1/4) Epoch 15, batch 4100, loss[loss=0.117, simple_loss=0.1489, pruned_loss=0.04254, over 14248.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1475, pruned_loss=0.03981, over 1875905.75 frames. ], batch size: 80, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:43:18,112 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:31,525 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.583e+01 2.156e+02 2.609e+02 3.160e+02 7.009e+02, threshold=5.218e+02, percent-clipped=5.0 2022-12-08 07:43:39,880 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:58,083 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6756, 5.1546, 5.0909, 5.6442, 5.2319, 4.7130, 5.6069, 4.6621], device='cuda:1'), covar=tensor([0.0332, 0.1017, 0.0353, 0.0397, 0.0787, 0.0378, 0.0492, 0.0500], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0271, 0.0197, 0.0194, 0.0185, 0.0154, 0.0286, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 07:43:58,956 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:00,684 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110017.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:05,364 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110022.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:18,996 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8627, 1.9427, 2.1189, 1.4121, 1.5322, 1.8590, 1.1864, 1.8476], device='cuda:1'), covar=tensor([0.1521, 0.1907, 0.0814, 0.2656, 0.2910, 0.1104, 0.3265, 0.1052], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0098, 0.0115, 0.0088, 0.0120, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 07:44:24,305 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110044.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:46,251 INFO [train.py:873] (1/4) Epoch 15, batch 4200, loss[loss=0.1483, simple_loss=0.1429, pruned_loss=0.07682, over 1268.00 frames. ], tot_loss[loss=0.113, simple_loss=0.1471, pruned_loss=0.03949, over 1916107.55 frames. ], batch size: 100, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:44:54,491 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:58,665 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2972, 2.4484, 4.9891, 4.4937, 4.2972, 5.1040, 4.7714, 5.0840], device='cuda:1'), covar=tensor([0.1385, 0.1158, 0.0069, 0.0185, 0.0206, 0.0097, 0.0119, 0.0097], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0128, 0.0168, 0.0145, 0.0140, 0.0122, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 07:45:03,731 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.176e+02 2.682e+02 3.403e+02 7.328e+02, threshold=5.365e+02, percent-clipped=6.0 2022-12-08 07:45:18,651 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110105.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:22,786 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:27,111 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:39,138 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7746, 2.4258, 3.8243, 3.9355, 3.7780, 2.2495, 3.8947, 2.9072], device='cuda:1'), covar=tensor([0.0465, 0.1165, 0.0877, 0.0485, 0.0477, 0.1848, 0.0463, 0.1039], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0257, 0.0372, 0.0329, 0.0270, 0.0303, 0.0309, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:45:55,057 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1353, 2.2202, 2.0339, 2.2586, 1.8760, 2.0145, 2.2104, 2.1824], device='cuda:1'), covar=tensor([0.0850, 0.1071, 0.1058, 0.0805, 0.1333, 0.1036, 0.0886, 0.0841], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0139, 0.0143, 0.0157, 0.0144, 0.0122, 0.0166, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:46:02,919 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4574, 1.8743, 3.5340, 2.5029, 3.4273, 1.8919, 2.7294, 3.2707], device='cuda:1'), covar=tensor([0.1040, 0.4476, 0.0758, 0.5679, 0.0896, 0.3766, 0.1482, 0.0740], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0208, 0.0216, 0.0279, 0.0234, 0.0208, 0.0209, 0.0218], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 07:46:13,001 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9042, 1.8482, 1.9244, 1.6333, 1.7324, 1.6532, 1.6860, 1.3615], device='cuda:1'), covar=tensor([0.0278, 0.0451, 0.0357, 0.0434, 0.0321, 0.0302, 0.0270, 0.0458], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0031, 0.0026, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:46:13,632 INFO [train.py:873] (1/4) Epoch 15, batch 4300, loss[loss=0.1004, simple_loss=0.1459, pruned_loss=0.02744, over 14270.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.147, pruned_loss=0.03906, over 1965396.81 frames. ], batch size: 57, lr: 5.24e-03, grad_scale: 4.0 2022-12-08 07:46:15,630 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:46:20,270 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:46:27,288 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8851, 2.9975, 4.6311, 3.4141, 4.5443, 4.4425, 4.3801, 3.9456], device='cuda:1'), covar=tensor([0.0883, 0.3415, 0.1061, 0.1869, 0.0964, 0.1069, 0.1516, 0.1583], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0313, 0.0389, 0.0299, 0.0371, 0.0320, 0.0359, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:46:32,112 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 2.281e+02 2.830e+02 3.301e+02 1.295e+03, threshold=5.661e+02, percent-clipped=5.0 2022-12-08 07:47:10,719 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3119, 1.9479, 3.4278, 2.4126, 3.2728, 1.9541, 2.6424, 3.2439], device='cuda:1'), covar=tensor([0.0910, 0.4220, 0.0648, 0.5315, 0.0825, 0.3273, 0.1477, 0.0697], device='cuda:1'), in_proj_covar=tensor([0.0254, 0.0206, 0.0213, 0.0276, 0.0232, 0.0206, 0.0207, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:47:12,398 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110235.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:47:32,270 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7633, 3.5123, 3.3038, 3.4364, 3.6997, 3.7129, 3.7662, 3.7335], device='cuda:1'), covar=tensor([0.0909, 0.0606, 0.2242, 0.2596, 0.0796, 0.0917, 0.0933, 0.0927], device='cuda:1'), in_proj_covar=tensor([0.0381, 0.0259, 0.0445, 0.0558, 0.0336, 0.0440, 0.0387, 0.0384], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:47:41,817 INFO [train.py:873] (1/4) Epoch 15, batch 4400, loss[loss=0.1185, simple_loss=0.1609, pruned_loss=0.03803, over 14061.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1481, pruned_loss=0.04015, over 1991032.13 frames. ], batch size: 29, lr: 5.24e-03, grad_scale: 8.0 2022-12-08 07:47:54,413 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:47:54,534 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:00,237 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.123e+01 2.204e+02 2.570e+02 3.174e+02 5.425e+02, threshold=5.140e+02, percent-clipped=0.0 2022-12-08 07:48:12,340 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110303.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:22,542 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:25,407 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:38,410 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6748, 2.5029, 3.1484, 2.0244, 2.0749, 2.7102, 1.5045, 2.7407], device='cuda:1'), covar=tensor([0.0925, 0.1194, 0.0651, 0.2384, 0.2050, 0.0866, 0.3192, 0.0936], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0098, 0.0115, 0.0088, 0.0120, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 07:48:40,170 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1261, 1.9839, 2.2193, 2.3049, 2.0232, 1.9494, 2.2922, 2.0685], device='cuda:1'), covar=tensor([0.0294, 0.0651, 0.0281, 0.0279, 0.0381, 0.0869, 0.0385, 0.0344], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0257, 0.0373, 0.0329, 0.0271, 0.0303, 0.0311, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:48:48,609 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110344.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:05,190 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:06,156 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:10,332 INFO [train.py:873] (1/4) Epoch 15, batch 4500, loss[loss=0.1553, simple_loss=0.1725, pruned_loss=0.06906, over 8629.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1473, pruned_loss=0.03915, over 2024398.00 frames. ], batch size: 100, lr: 5.24e-03, grad_scale: 8.0 2022-12-08 07:49:14,193 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110373.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:20,002 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110379.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:29,133 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.023e+02 2.685e+02 3.179e+02 5.561e+02, threshold=5.371e+02, percent-clipped=3.0 2022-12-08 07:49:38,282 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110400.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:50,761 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4797, 1.9375, 2.4732, 2.0870, 2.4507, 2.4056, 2.2945, 2.2261], device='cuda:1'), covar=tensor([0.0579, 0.2537, 0.0901, 0.1540, 0.0626, 0.1099, 0.0630, 0.1345], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0315, 0.0392, 0.0299, 0.0373, 0.0321, 0.0360, 0.0301], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:50:13,545 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5344, 1.4336, 2.7423, 1.4997, 2.7393, 2.7099, 2.0562, 2.8448], device='cuda:1'), covar=tensor([0.0343, 0.2910, 0.0479, 0.2107, 0.0460, 0.0571, 0.1106, 0.0354], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0160, 0.0170, 0.0168, 0.0179, 0.0133, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:50:36,591 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:50:39,116 INFO [train.py:873] (1/4) Epoch 15, batch 4600, loss[loss=0.1313, simple_loss=0.156, pruned_loss=0.05333, over 7803.00 frames. ], tot_loss[loss=0.1149, simple_loss=0.1484, pruned_loss=0.04069, over 1985184.96 frames. ], batch size: 100, lr: 5.24e-03, grad_scale: 4.0 2022-12-08 07:50:40,932 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:50:48,628 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.69 vs. limit=5.0 2022-12-08 07:50:58,403 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.355e+02 2.918e+02 3.425e+02 9.705e+02, threshold=5.836e+02, percent-clipped=5.0 2022-12-08 07:51:22,851 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-08 07:51:32,412 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7866, 2.7713, 2.5922, 2.8757, 2.4605, 2.5355, 2.8258, 2.7572], device='cuda:1'), covar=tensor([0.0780, 0.1060, 0.0969, 0.0761, 0.1206, 0.0896, 0.0827, 0.0876], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0140, 0.0145, 0.0158, 0.0145, 0.0123, 0.0168, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 07:52:06,647 INFO [train.py:873] (1/4) Epoch 15, batch 4700, loss[loss=0.1255, simple_loss=0.1541, pruned_loss=0.04845, over 5956.00 frames. ], tot_loss[loss=0.1148, simple_loss=0.1481, pruned_loss=0.04078, over 1983426.64 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 4.0 2022-12-08 07:52:26,207 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.034e+02 2.470e+02 3.158e+02 1.519e+03, threshold=4.939e+02, percent-clipped=4.0 2022-12-08 07:52:44,559 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110611.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:05,031 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 07:53:09,081 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110639.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:26,649 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:34,854 INFO [train.py:873] (1/4) Epoch 15, batch 4800, loss[loss=0.1555, simple_loss=0.1763, pruned_loss=0.06734, over 8603.00 frames. ], tot_loss[loss=0.1132, simple_loss=0.1472, pruned_loss=0.03957, over 2000698.79 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 8.0 2022-12-08 07:53:37,791 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110672.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:38,585 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110673.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:39,419 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110674.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:53,744 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.088e+02 2.309e+02 2.792e+02 3.592e+02 7.510e+02, threshold=5.583e+02, percent-clipped=5.0 2022-12-08 07:53:53,928 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1992, 1.1586, 1.0731, 1.1698, 1.1933, 0.7417, 0.9830, 1.1878], device='cuda:1'), covar=tensor([0.0638, 0.0802, 0.0644, 0.0708, 0.0620, 0.0789, 0.1108, 0.0787], device='cuda:1'), in_proj_covar=tensor([0.0033, 0.0032, 0.0036, 0.0030, 0.0032, 0.0046, 0.0033, 0.0036], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:54:01,995 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110700.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:19,716 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9734, 2.6086, 3.4624, 2.4033, 2.1059, 3.0093, 1.7229, 2.9616], device='cuda:1'), covar=tensor([0.1222, 0.1323, 0.0793, 0.2020, 0.2283, 0.0955, 0.3279, 0.1077], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0098, 0.0115, 0.0088, 0.0119, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 07:54:20,437 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110721.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:33,584 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:44,445 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110748.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:00,677 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:03,180 INFO [train.py:873] (1/4) Epoch 15, batch 4900, loss[loss=0.1511, simple_loss=0.147, pruned_loss=0.07764, over 1251.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.1477, pruned_loss=0.03991, over 2021728.66 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 8.0 2022-12-08 07:55:04,851 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:22,713 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 1.962e+02 2.570e+02 3.049e+02 5.982e+02, threshold=5.139e+02, percent-clipped=2.0 2022-12-08 07:55:26,127 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 07:55:28,504 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110797.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:55:43,512 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:48,026 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:56:20,513 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 07:56:31,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110868.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:56:31,858 INFO [train.py:873] (1/4) Epoch 15, batch 5000, loss[loss=0.1061, simple_loss=0.1398, pruned_loss=0.03618, over 14156.00 frames. ], tot_loss[loss=0.114, simple_loss=0.148, pruned_loss=0.04002, over 2061680.81 frames. ], batch size: 99, lr: 5.23e-03, grad_scale: 4.0 2022-12-08 07:56:35,761 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-08 07:56:52,356 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.709e+01 2.105e+02 2.624e+02 3.225e+02 7.187e+02, threshold=5.248e+02, percent-clipped=2.0 2022-12-08 07:57:25,019 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:31,867 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-08 07:57:33,759 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:46,482 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7747, 1.9877, 2.6665, 2.2122, 2.6028, 2.6321, 2.4730, 2.4299], device='cuda:1'), covar=tensor([0.0812, 0.2720, 0.0783, 0.1681, 0.0662, 0.0978, 0.0880, 0.1387], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0318, 0.0395, 0.0302, 0.0378, 0.0327, 0.0363, 0.0304], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:57:49,686 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2022-12-08 07:57:50,984 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110958.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:51,699 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110959.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:57,687 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8447, 1.6485, 2.0884, 1.6108, 1.8203, 1.4280, 1.6755, 1.9288], device='cuda:1'), covar=tensor([0.2888, 0.2444, 0.0562, 0.1564, 0.1353, 0.1117, 0.0987, 0.0639], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0205, 0.0213, 0.0276, 0.0231, 0.0207, 0.0206, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:57:58,601 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110967.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:00,305 INFO [train.py:873] (1/4) Epoch 15, batch 5100, loss[loss=0.1171, simple_loss=0.1476, pruned_loss=0.04334, over 14241.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1476, pruned_loss=0.03971, over 1979673.30 frames. ], batch size: 46, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 07:58:04,951 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:16,279 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:20,532 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.198e+02 2.177e+02 2.703e+02 3.385e+02 6.811e+02, threshold=5.407e+02, percent-clipped=6.0 2022-12-08 07:58:33,413 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:44,158 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111019.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:46,637 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111022.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:58,051 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5348, 1.6202, 1.6619, 1.3814, 1.3926, 1.4284, 1.4174, 1.2037], device='cuda:1'), covar=tensor([0.0174, 0.0196, 0.0130, 0.0205, 0.0183, 0.0310, 0.0203, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:59:26,199 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 07:59:28,410 INFO [train.py:873] (1/4) Epoch 15, batch 5200, loss[loss=0.1222, simple_loss=0.1524, pruned_loss=0.04595, over 14220.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.1479, pruned_loss=0.03982, over 2005285.80 frames. ], batch size: 89, lr: 5.22e-03, grad_scale: 8.0 2022-12-08 07:59:33,495 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0775, 1.8266, 1.8502, 2.0030, 1.9134, 1.0720, 1.7845, 1.7879], device='cuda:1'), covar=tensor([0.0799, 0.0658, 0.0882, 0.0540, 0.0697, 0.1036, 0.0845, 0.1385], device='cuda:1'), in_proj_covar=tensor([0.0034, 0.0033, 0.0036, 0.0031, 0.0032, 0.0047, 0.0033, 0.0037], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 07:59:39,915 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2040, 2.1632, 3.0872, 2.3916, 3.1192, 3.0481, 2.9302, 2.6363], device='cuda:1'), covar=tensor([0.1347, 0.3240, 0.1075, 0.2032, 0.0784, 0.0980, 0.1097, 0.1707], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0313, 0.0391, 0.0296, 0.0373, 0.0322, 0.0358, 0.0300], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 07:59:48,545 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.014e+01 2.007e+02 2.591e+02 3.079e+02 5.022e+02, threshold=5.182e+02, percent-clipped=0.0 2022-12-08 07:59:48,689 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111092.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 08:00:00,454 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7939, 1.3253, 2.4580, 2.1887, 2.3676, 2.5152, 1.6391, 2.4690], device='cuda:1'), covar=tensor([0.1144, 0.1612, 0.0259, 0.0556, 0.0560, 0.0288, 0.0847, 0.0293], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0157, 0.0128, 0.0168, 0.0145, 0.0140, 0.0121, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 08:00:35,807 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 08:00:57,147 INFO [train.py:873] (1/4) Epoch 15, batch 5300, loss[loss=0.11, simple_loss=0.1462, pruned_loss=0.03687, over 14163.00 frames. ], tot_loss[loss=0.1132, simple_loss=0.1474, pruned_loss=0.0395, over 2017619.42 frames. ], batch size: 99, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 08:00:58,425 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2022-12-08 08:00:59,054 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9716, 1.9953, 3.9878, 2.7082, 3.8191, 1.9255, 3.0860, 3.7911], device='cuda:1'), covar=tensor([0.0630, 0.4349, 0.0536, 0.5641, 0.0679, 0.3647, 0.1428, 0.0466], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0206, 0.0214, 0.0277, 0.0231, 0.0207, 0.0207, 0.0218], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 08:01:18,476 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.026e+02 2.377e+02 2.924e+02 1.186e+03, threshold=4.754e+02, percent-clipped=4.0 2022-12-08 08:01:46,191 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111224.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:01:46,999 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:24,027 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111267.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:25,564 INFO [train.py:873] (1/4) Epoch 15, batch 5400, loss[loss=0.1929, simple_loss=0.169, pruned_loss=0.1084, over 1260.00 frames. ], tot_loss[loss=0.1129, simple_loss=0.1477, pruned_loss=0.03905, over 2104354.61 frames. ], batch size: 100, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 08:02:40,937 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:46,809 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.071e+02 2.546e+02 3.124e+02 5.888e+02, threshold=5.092e+02, percent-clipped=1.0 2022-12-08 08:03:05,564 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111314.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:03:06,444 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:03:23,839 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2004, 2.0381, 2.1447, 2.2243, 2.1180, 2.1148, 2.2561, 1.9245], device='cuda:1'), covar=tensor([0.0920, 0.1423, 0.0762, 0.0785, 0.1098, 0.0769, 0.0951, 0.0735], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0271, 0.0198, 0.0195, 0.0186, 0.0156, 0.0286, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:03:53,189 INFO [train.py:873] (1/4) Epoch 15, batch 5500, loss[loss=0.1096, simple_loss=0.1164, pruned_loss=0.0514, over 2568.00 frames. ], tot_loss[loss=0.1119, simple_loss=0.1467, pruned_loss=0.03852, over 2056883.00 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:04:13,895 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111392.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:04:14,624 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.065e+02 2.464e+02 3.206e+02 5.854e+02, threshold=4.928e+02, percent-clipped=3.0 2022-12-08 08:04:29,543 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3173, 3.0399, 2.9990, 3.2778, 3.1320, 3.2564, 3.3329, 2.7119], device='cuda:1'), covar=tensor([0.0563, 0.1090, 0.0580, 0.0517, 0.0872, 0.0479, 0.0636, 0.0692], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0273, 0.0198, 0.0196, 0.0188, 0.0156, 0.0288, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:04:56,367 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:05:20,887 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:05:21,569 INFO [train.py:873] (1/4) Epoch 15, batch 5600, loss[loss=0.1312, simple_loss=0.1559, pruned_loss=0.05324, over 9485.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1468, pruned_loss=0.03885, over 2024882.33 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 8.0 2022-12-08 08:05:31,635 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8484, 1.6040, 1.8134, 1.9882, 1.4483, 1.6740, 1.6928, 1.8760], device='cuda:1'), covar=tensor([0.0218, 0.0363, 0.0200, 0.0194, 0.0367, 0.0419, 0.0280, 0.0192], device='cuda:1'), in_proj_covar=tensor([0.0291, 0.0256, 0.0371, 0.0326, 0.0270, 0.0304, 0.0309, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:05:42,278 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.013e+02 2.578e+02 3.220e+02 8.105e+02, threshold=5.156e+02, percent-clipped=4.0 2022-12-08 08:05:52,487 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2823, 2.9386, 5.2166, 3.7874, 5.0217, 2.7427, 4.0488, 4.9021], device='cuda:1'), covar=tensor([0.0403, 0.3729, 0.0410, 0.5079, 0.0450, 0.2961, 0.1102, 0.0339], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0208, 0.0217, 0.0279, 0.0234, 0.0208, 0.0209, 0.0219], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 08:06:07,189 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 08:06:10,265 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:14,555 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:48,893 INFO [train.py:873] (1/4) Epoch 15, batch 5700, loss[loss=0.1153, simple_loss=0.1549, pruned_loss=0.03784, over 14114.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1471, pruned_loss=0.03878, over 2042748.82 frames. ], batch size: 29, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:06:51,464 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111572.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:59,441 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111581.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:07:08,229 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6783, 1.7195, 2.8512, 2.1586, 2.7192, 1.7198, 2.2517, 2.6277], device='cuda:1'), covar=tensor([0.1344, 0.3443, 0.0669, 0.3463, 0.0999, 0.2876, 0.1260, 0.0796], device='cuda:1'), in_proj_covar=tensor([0.0255, 0.0206, 0.0216, 0.0277, 0.0232, 0.0206, 0.0207, 0.0217], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:07:10,365 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.509e+01 2.357e+02 2.792e+02 3.438e+02 6.748e+02, threshold=5.583e+02, percent-clipped=1.0 2022-12-08 08:07:28,749 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111614.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:08:10,715 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111662.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:08:11,628 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5898, 1.2064, 2.0230, 1.7984, 1.8227, 2.0803, 1.3116, 2.0741], device='cuda:1'), covar=tensor([0.0842, 0.1345, 0.0286, 0.0534, 0.0664, 0.0283, 0.0873, 0.0276], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0130, 0.0170, 0.0146, 0.0141, 0.0122, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 08:08:17,295 INFO [train.py:873] (1/4) Epoch 15, batch 5800, loss[loss=0.1121, simple_loss=0.145, pruned_loss=0.03956, over 14215.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.1478, pruned_loss=0.03994, over 1990108.14 frames. ], batch size: 37, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:08:38,925 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.211e+02 2.861e+02 3.266e+02 8.310e+02, threshold=5.722e+02, percent-clipped=4.0 2022-12-08 08:09:44,962 INFO [train.py:873] (1/4) Epoch 15, batch 5900, loss[loss=0.1516, simple_loss=0.1446, pruned_loss=0.07928, over 1221.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1471, pruned_loss=0.03981, over 1921879.69 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:10:04,801 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111791.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:10:07,183 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.002e+02 2.082e+02 2.433e+02 3.039e+02 5.689e+02, threshold=4.867e+02, percent-clipped=0.0 2022-12-08 08:10:17,408 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2667, 2.6375, 4.0063, 2.9876, 4.0991, 3.8611, 3.8454, 3.4826], device='cuda:1'), covar=tensor([0.0932, 0.3447, 0.1320, 0.2060, 0.0889, 0.1210, 0.1612, 0.1830], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0315, 0.0395, 0.0301, 0.0374, 0.0327, 0.0363, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:10:33,516 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:10:45,874 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-08 08:10:46,280 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8085, 4.4255, 4.2639, 4.7765, 4.4539, 4.2005, 4.7860, 3.9804], device='cuda:1'), covar=tensor([0.0362, 0.0859, 0.0432, 0.0391, 0.0745, 0.0645, 0.0475, 0.0510], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0269, 0.0195, 0.0193, 0.0185, 0.0155, 0.0282, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:10:57,474 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111852.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:11:12,396 INFO [train.py:873] (1/4) Epoch 15, batch 6000, loss[loss=0.08412, simple_loss=0.1257, pruned_loss=0.02129, over 13944.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.147, pruned_loss=0.04003, over 1911396.88 frames. ], batch size: 20, lr: 5.20e-03, grad_scale: 8.0 2022-12-08 08:11:12,397 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 08:11:16,524 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6564, 3.8204, 3.6607, 3.0646, 2.7642, 3.8346, 3.4830, 2.4082], device='cuda:1'), covar=tensor([0.1043, 0.0453, 0.0855, 0.0551, 0.0869, 0.0282, 0.0841, 0.1495], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0088, 0.0068, 0.0072, 0.0098, 0.0086, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:11:20,187 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4654, 3.3723, 3.6837, 2.9356, 2.3446, 3.1595, 1.8713, 3.0540], device='cuda:1'), covar=tensor([0.0305, 0.0298, 0.0279, 0.0944, 0.1730, 0.0651, 0.2480, 0.0617], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0102, 0.0094, 0.0099, 0.0116, 0.0090, 0.0121, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 08:11:20,777 INFO [train.py:905] (1/4) Epoch 15, validation: loss=0.1363, simple_loss=0.1737, pruned_loss=0.04946, over 857387.00 frames. 2022-12-08 08:11:20,778 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 08:11:31,614 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:11:42,936 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.150e+02 2.606e+02 3.231e+02 6.699e+02, threshold=5.212e+02, percent-clipped=6.0 2022-12-08 08:11:58,712 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-08 08:12:04,026 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 08:12:04,498 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7806, 1.8588, 1.5275, 1.9324, 1.7815, 1.7111, 1.7783, 1.6430], device='cuda:1'), covar=tensor([0.1382, 0.1064, 0.2392, 0.0818, 0.1238, 0.0682, 0.1880, 0.1080], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0291, 0.0259, 0.0280, 0.0323, 0.0301, 0.0255, 0.0246], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:12:11,310 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2162, 1.3791, 3.2792, 1.4919, 2.9862, 3.3398, 2.4416, 3.5288], device='cuda:1'), covar=tensor([0.0306, 0.3504, 0.0451, 0.2599, 0.1353, 0.0454, 0.0935, 0.0271], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0158, 0.0160, 0.0170, 0.0169, 0.0180, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:12:13,664 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:12:48,569 INFO [train.py:873] (1/4) Epoch 15, batch 6100, loss[loss=0.1326, simple_loss=0.1429, pruned_loss=0.0612, over 3867.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.147, pruned_loss=0.04027, over 1870751.39 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:13:10,956 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.212e+02 2.710e+02 3.481e+02 4.963e+02, threshold=5.420e+02, percent-clipped=0.0 2022-12-08 08:13:12,820 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 08:14:13,818 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:14:16,350 INFO [train.py:873] (1/4) Epoch 15, batch 6200, loss[loss=0.109, simple_loss=0.1534, pruned_loss=0.03226, over 14023.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1472, pruned_loss=0.03993, over 1940005.19 frames. ], batch size: 22, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:14:29,581 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7874, 1.2120, 2.5404, 2.2781, 2.4082, 2.5650, 1.7368, 2.5314], device='cuda:1'), covar=tensor([0.1008, 0.1501, 0.0237, 0.0480, 0.0461, 0.0251, 0.0818, 0.0285], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0130, 0.0171, 0.0146, 0.0142, 0.0123, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 08:14:39,268 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 2.103e+02 2.700e+02 3.313e+02 6.071e+02, threshold=5.399e+02, percent-clipped=3.0 2022-12-08 08:14:51,357 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2022-12-08 08:15:05,096 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:15:07,695 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:15:19,745 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8366, 1.8233, 1.6219, 1.9799, 1.8064, 1.7874, 1.8196, 1.7036], device='cuda:1'), covar=tensor([0.0943, 0.0992, 0.1691, 0.0687, 0.0952, 0.0569, 0.1454, 0.1057], device='cuda:1'), in_proj_covar=tensor([0.0276, 0.0290, 0.0257, 0.0279, 0.0321, 0.0300, 0.0253, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:15:25,310 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112147.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:15:44,577 INFO [train.py:873] (1/4) Epoch 15, batch 6300, loss[loss=0.1272, simple_loss=0.1259, pruned_loss=0.06428, over 1237.00 frames. ], tot_loss[loss=0.1127, simple_loss=0.1464, pruned_loss=0.03949, over 1900300.48 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:15:47,428 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:16:07,628 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.167e+02 2.529e+02 3.043e+02 7.690e+02, threshold=5.057e+02, percent-clipped=2.0 2022-12-08 08:16:22,013 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7923, 2.8568, 2.9667, 2.8632, 2.8571, 2.6514, 1.5674, 2.6146], device='cuda:1'), covar=tensor([0.0464, 0.0413, 0.0397, 0.0436, 0.0400, 0.1004, 0.2623, 0.0410], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0174, 0.0144, 0.0146, 0.0206, 0.0141, 0.0160, 0.0195], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 08:16:25,126 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7529, 2.3596, 3.5206, 2.6451, 3.5837, 3.4956, 3.3518, 2.9193], device='cuda:1'), covar=tensor([0.0807, 0.2933, 0.1150, 0.1940, 0.0747, 0.0969, 0.1176, 0.1828], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0313, 0.0396, 0.0300, 0.0371, 0.0327, 0.0364, 0.0302], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:16:31,079 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8954, 1.5659, 1.8126, 1.9303, 1.4637, 1.7023, 1.8374, 1.7859], device='cuda:1'), covar=tensor([0.0235, 0.0372, 0.0197, 0.0184, 0.0459, 0.0398, 0.0268, 0.0194], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0261, 0.0375, 0.0330, 0.0272, 0.0307, 0.0314, 0.0283], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:16:34,129 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 08:16:37,507 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1180, 4.6607, 4.5319, 5.0711, 4.5726, 4.4482, 5.0417, 4.2181], device='cuda:1'), covar=tensor([0.0339, 0.1053, 0.0376, 0.0416, 0.1008, 0.0604, 0.0540, 0.0559], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0270, 0.0195, 0.0191, 0.0185, 0.0155, 0.0281, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:17:13,384 INFO [train.py:873] (1/4) Epoch 15, batch 6400, loss[loss=0.1282, simple_loss=0.1523, pruned_loss=0.05205, over 10346.00 frames. ], tot_loss[loss=0.1124, simple_loss=0.1465, pruned_loss=0.03915, over 1944655.89 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:17:35,982 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 2.167e+02 2.735e+02 3.656e+02 1.132e+03, threshold=5.470e+02, percent-clipped=11.0 2022-12-08 08:17:45,446 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5395, 3.8559, 3.7864, 3.4504, 2.7663, 3.7219, 3.5646, 2.1121], device='cuda:1'), covar=tensor([0.1332, 0.0737, 0.0516, 0.1052, 0.0907, 0.0521, 0.0920, 0.1925], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0088, 0.0068, 0.0071, 0.0097, 0.0086, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:17:56,915 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1067, 1.0313, 1.1044, 0.9592, 1.0089, 0.8206, 0.9266, 0.8417], device='cuda:1'), covar=tensor([0.0255, 0.0205, 0.0187, 0.0228, 0.0225, 0.0370, 0.0250, 0.0365], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:18:41,068 INFO [train.py:873] (1/4) Epoch 15, batch 6500, loss[loss=0.1215, simple_loss=0.1447, pruned_loss=0.04913, over 4997.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1471, pruned_loss=0.03983, over 1948485.55 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:18:59,385 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112389.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:19:04,118 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.163e+02 2.578e+02 3.269e+02 5.158e+02, threshold=5.155e+02, percent-clipped=0.0 2022-12-08 08:19:28,060 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112422.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:19:50,533 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112447.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:19:53,052 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:20:09,726 INFO [train.py:873] (1/4) Epoch 15, batch 6600, loss[loss=0.1298, simple_loss=0.1603, pruned_loss=0.04966, over 10319.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1466, pruned_loss=0.03924, over 1965334.38 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:20:29,861 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2022-12-08 08:20:32,876 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.043e+02 2.558e+02 3.259e+02 6.600e+02, threshold=5.115e+02, percent-clipped=3.0 2022-12-08 08:20:32,964 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112495.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:20:59,978 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0375, 1.2351, 1.3289, 1.0209, 0.8582, 1.1728, 0.9113, 1.2236], device='cuda:1'), covar=tensor([0.1902, 0.2731, 0.1000, 0.2438, 0.3210, 0.1357, 0.1641, 0.1197], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0100, 0.0093, 0.0099, 0.0115, 0.0088, 0.0119, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 08:21:13,313 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4673, 2.2508, 3.3789, 3.5707, 3.3906, 2.1683, 3.4600, 2.5887], device='cuda:1'), covar=tensor([0.0483, 0.1214, 0.0859, 0.0497, 0.0590, 0.1721, 0.0470, 0.1084], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0259, 0.0374, 0.0329, 0.0271, 0.0305, 0.0312, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:21:32,141 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112562.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:21:37,740 INFO [train.py:873] (1/4) Epoch 15, batch 6700, loss[loss=0.1245, simple_loss=0.1463, pruned_loss=0.05129, over 6050.00 frames. ], tot_loss[loss=0.1127, simple_loss=0.147, pruned_loss=0.03925, over 1906107.59 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:21:44,977 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:22:01,001 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.231e+02 2.690e+02 3.411e+02 7.258e+02, threshold=5.380e+02, percent-clipped=8.0 2022-12-08 08:22:20,704 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:22:26,096 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112623.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:22:39,502 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112638.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 08:23:06,602 INFO [train.py:873] (1/4) Epoch 15, batch 6800, loss[loss=0.1065, simple_loss=0.1456, pruned_loss=0.03373, over 14264.00 frames. ], tot_loss[loss=0.1131, simple_loss=0.1471, pruned_loss=0.03953, over 1921578.62 frames. ], batch size: 57, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:23:14,271 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112678.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:23:21,147 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6419, 1.8133, 1.8706, 1.8455, 1.8617, 1.6291, 1.5140, 1.2149], device='cuda:1'), covar=tensor([0.0311, 0.0274, 0.0294, 0.0247, 0.0208, 0.0321, 0.0293, 0.0501], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:23:28,859 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.249e+02 2.719e+02 3.532e+02 9.106e+02, threshold=5.438e+02, percent-clipped=9.0 2022-12-08 08:23:34,200 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6934, 1.7113, 2.8930, 2.2683, 2.7335, 1.8409, 2.3158, 2.6643], device='cuda:1'), covar=tensor([0.1400, 0.3732, 0.0763, 0.3420, 0.1112, 0.3036, 0.1170, 0.0769], device='cuda:1'), in_proj_covar=tensor([0.0255, 0.0204, 0.0216, 0.0276, 0.0233, 0.0209, 0.0208, 0.0217], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 08:23:52,910 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112722.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:23:57,746 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-08 08:24:13,175 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112745.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:24:34,754 INFO [train.py:873] (1/4) Epoch 15, batch 6900, loss[loss=0.09428, simple_loss=0.1401, pruned_loss=0.02422, over 14081.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1469, pruned_loss=0.03988, over 1899689.29 frames. ], batch size: 29, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:24:35,716 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:24:57,284 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.169e+02 2.514e+02 3.179e+02 8.231e+02, threshold=5.028e+02, percent-clipped=2.0 2022-12-08 08:26:02,334 INFO [train.py:873] (1/4) Epoch 15, batch 7000, loss[loss=0.1392, simple_loss=0.1684, pruned_loss=0.055, over 14355.00 frames. ], tot_loss[loss=0.1131, simple_loss=0.1468, pruned_loss=0.03972, over 1860363.51 frames. ], batch size: 73, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:26:07,154 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8390, 1.6045, 2.0644, 1.6802, 1.9101, 1.4007, 1.6938, 1.9536], device='cuda:1'), covar=tensor([0.2913, 0.2507, 0.0656, 0.1946, 0.1447, 0.1594, 0.0989, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0202, 0.0213, 0.0274, 0.0231, 0.0207, 0.0205, 0.0215], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:26:25,455 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.195e+02 2.715e+02 3.350e+02 5.262e+02, threshold=5.431e+02, percent-clipped=3.0 2022-12-08 08:26:39,526 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 08:26:46,181 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:26:49,271 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112921.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:26:59,618 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112933.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:27:31,513 INFO [train.py:873] (1/4) Epoch 15, batch 7100, loss[loss=0.1191, simple_loss=0.1567, pruned_loss=0.04078, over 14386.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1465, pruned_loss=0.03906, over 1858730.54 frames. ], batch size: 55, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:27:33,611 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 08:27:35,246 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112973.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 08:27:43,040 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112982.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:27:54,134 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.181e+02 2.817e+02 3.355e+02 7.638e+02, threshold=5.634e+02, percent-clipped=2.0 2022-12-08 08:28:17,410 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1514, 1.4873, 3.2215, 1.6169, 3.1509, 3.2277, 2.3825, 3.4366], device='cuda:1'), covar=tensor([0.0242, 0.2770, 0.0465, 0.2087, 0.0840, 0.0461, 0.0991, 0.0233], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0156, 0.0160, 0.0170, 0.0169, 0.0180, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:28:34,742 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 08:28:38,840 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113045.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:28:59,911 INFO [train.py:873] (1/4) Epoch 15, batch 7200, loss[loss=0.1084, simple_loss=0.1436, pruned_loss=0.03658, over 14163.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1459, pruned_loss=0.03837, over 1931077.64 frames. ], batch size: 99, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:29:21,092 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113093.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:29:23,117 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.228e+02 2.733e+02 3.389e+02 6.488e+02, threshold=5.466e+02, percent-clipped=1.0 2022-12-08 08:30:28,628 INFO [train.py:873] (1/4) Epoch 15, batch 7300, loss[loss=0.1331, simple_loss=0.1393, pruned_loss=0.06347, over 2658.00 frames. ], tot_loss[loss=0.1116, simple_loss=0.1459, pruned_loss=0.03859, over 1920392.13 frames. ], batch size: 100, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:30:28,706 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7837, 2.6542, 2.2476, 2.4412, 2.7644, 2.7203, 2.7507, 2.7321], device='cuda:1'), covar=tensor([0.1378, 0.1023, 0.3220, 0.3306, 0.1390, 0.1670, 0.1785, 0.1386], device='cuda:1'), in_proj_covar=tensor([0.0383, 0.0263, 0.0440, 0.0558, 0.0341, 0.0438, 0.0387, 0.0385], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:30:52,685 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.057e+02 2.452e+02 3.236e+02 1.011e+03, threshold=4.903e+02, percent-clipped=3.0 2022-12-08 08:31:12,614 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:15,390 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2600, 2.0890, 2.1434, 2.0225, 2.0800, 1.3584, 2.0611, 2.3129], device='cuda:1'), covar=tensor([0.0906, 0.0982, 0.0526, 0.1491, 0.0675, 0.0878, 0.0762, 0.0593], device='cuda:1'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0030, 0.0032, 0.0046, 0.0034, 0.0036], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:31:24,388 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 08:31:25,885 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:26,717 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:55,296 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113266.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:57,792 INFO [train.py:873] (1/4) Epoch 15, batch 7400, loss[loss=0.1145, simple_loss=0.1516, pruned_loss=0.03873, over 14439.00 frames. ], tot_loss[loss=0.1121, simple_loss=0.1463, pruned_loss=0.03892, over 1961644.71 frames. ], batch size: 53, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:32:01,637 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113273.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 08:32:05,495 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113277.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:08,855 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:21,745 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113295.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:22,353 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.108e+02 1.996e+02 2.516e+02 3.094e+02 5.431e+02, threshold=5.033e+02, percent-clipped=1.0 2022-12-08 08:32:44,277 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:44,347 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8946, 1.9484, 2.0838, 1.4666, 1.4858, 1.9305, 1.3074, 1.9347], device='cuda:1'), covar=tensor([0.1206, 0.1957, 0.0982, 0.2397, 0.2794, 0.1151, 0.3675, 0.1050], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0101, 0.0092, 0.0099, 0.0116, 0.0088, 0.0120, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 08:32:46,158 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2329, 1.3691, 4.0163, 1.8710, 4.0875, 4.2968, 3.5196, 4.6405], device='cuda:1'), covar=tensor([0.0235, 0.3473, 0.0453, 0.2274, 0.0350, 0.0409, 0.0562, 0.0153], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0158, 0.0161, 0.0172, 0.0170, 0.0181, 0.0135, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:33:25,700 INFO [train.py:873] (1/4) Epoch 15, batch 7500, loss[loss=0.09516, simple_loss=0.138, pruned_loss=0.02615, over 14632.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1468, pruned_loss=0.03937, over 1975279.20 frames. ], batch size: 22, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:33:48,619 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.151e+02 2.525e+02 3.725e+02 6.084e+02, threshold=5.049e+02, percent-clipped=3.0 2022-12-08 08:34:01,320 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2022-12-08 08:34:54,699 INFO [train.py:873] (1/4) Epoch 16, batch 0, loss[loss=0.1375, simple_loss=0.1212, pruned_loss=0.07689, over 1278.00 frames. ], tot_loss[loss=0.1375, simple_loss=0.1212, pruned_loss=0.07689, over 1278.00 frames. ], batch size: 100, lr: 5.00e-03, grad_scale: 8.0 2022-12-08 08:34:54,699 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 08:34:59,812 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9289, 1.6997, 1.7704, 1.7483, 1.7014, 1.6543, 1.4440, 1.5321], device='cuda:1'), covar=tensor([0.0140, 0.0254, 0.0161, 0.0157, 0.0158, 0.0267, 0.0188, 0.0239], device='cuda:1'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0031, 0.0026, 0.0030], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:35:01,997 INFO [train.py:905] (1/4) Epoch 16, validation: loss=0.1445, simple_loss=0.1858, pruned_loss=0.05158, over 857387.00 frames. 2022-12-08 08:35:01,997 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 08:35:03,905 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6583, 1.8170, 3.7040, 2.5376, 3.5661, 1.8086, 2.6692, 3.4896], device='cuda:1'), covar=tensor([0.0711, 0.4447, 0.0549, 0.5498, 0.0960, 0.3776, 0.1691, 0.0693], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0204, 0.0215, 0.0275, 0.0232, 0.0206, 0.0207, 0.0215], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:35:16,561 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9674, 3.4339, 3.2218, 3.3648, 2.4801, 3.4123, 3.1728, 1.6579], device='cuda:1'), covar=tensor([0.1688, 0.0689, 0.1241, 0.0570, 0.1008, 0.0569, 0.1130, 0.2190], device='cuda:1'), in_proj_covar=tensor([0.0142, 0.0088, 0.0068, 0.0072, 0.0098, 0.0086, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:35:30,757 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8510, 1.5185, 3.7206, 3.3466, 3.4979, 3.7666, 3.1513, 3.7272], device='cuda:1'), covar=tensor([0.1605, 0.1686, 0.0135, 0.0306, 0.0286, 0.0157, 0.0271, 0.0155], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0160, 0.0132, 0.0171, 0.0148, 0.0143, 0.0125, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 08:36:01,324 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.044e+01 2.013e+02 2.778e+02 4.051e+02 1.103e+03, threshold=5.556e+02, percent-clipped=15.0 2022-12-08 08:36:21,192 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7927, 3.5625, 3.3778, 2.6276, 3.1963, 3.6050, 3.8388, 3.1507], device='cuda:1'), covar=tensor([0.0493, 0.0923, 0.0731, 0.1123, 0.0907, 0.0602, 0.0596, 0.0963], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0172, 0.0137, 0.0125, 0.0141, 0.0153, 0.0130, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:36:31,855 INFO [train.py:873] (1/4) Epoch 16, batch 100, loss[loss=0.1094, simple_loss=0.1486, pruned_loss=0.03505, over 14097.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.1467, pruned_loss=0.03775, over 857540.86 frames. ], batch size: 29, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:37:13,536 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:25,098 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113590.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:31,129 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.137e+02 2.757e+02 3.201e+02 6.700e+02, threshold=5.514e+02, percent-clipped=1.0 2022-12-08 08:37:49,467 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:56,337 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:00,188 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6408, 2.7722, 4.4061, 3.3447, 4.3966, 4.2291, 4.1236, 3.8492], device='cuda:1'), covar=tensor([0.0563, 0.3054, 0.0722, 0.1641, 0.0650, 0.0841, 0.1449, 0.1664], device='cuda:1'), in_proj_covar=tensor([0.0347, 0.0308, 0.0387, 0.0293, 0.0365, 0.0318, 0.0355, 0.0297], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:38:01,673 INFO [train.py:873] (1/4) Epoch 16, batch 200, loss[loss=0.117, simple_loss=0.1283, pruned_loss=0.05288, over 2594.00 frames. ], tot_loss[loss=0.1103, simple_loss=0.1457, pruned_loss=0.03739, over 1343997.46 frames. ], batch size: 100, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:38:14,897 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:19,809 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2022-12-08 08:38:31,678 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8358, 1.2897, 2.0490, 1.2237, 1.9413, 2.0159, 1.7142, 2.1034], device='cuda:1'), covar=tensor([0.0360, 0.2181, 0.0508, 0.1963, 0.0609, 0.0693, 0.1093, 0.0432], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0159, 0.0168, 0.0167, 0.0178, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:38:43,101 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:59,610 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.233e+02 2.172e+02 2.621e+02 3.364e+02 2.345e+03, threshold=5.243e+02, percent-clipped=6.0 2022-12-08 08:39:02,871 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8206, 2.4122, 3.1688, 1.9920, 2.0323, 2.6525, 1.4871, 2.7903], device='cuda:1'), covar=tensor([0.0817, 0.1360, 0.0475, 0.2088, 0.2059, 0.0788, 0.3174, 0.0840], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0099, 0.0115, 0.0088, 0.0120, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 08:39:09,152 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:39:13,393 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3375, 2.0506, 2.3342, 2.3995, 2.2039, 1.9430, 2.3534, 2.1612], device='cuda:1'), covar=tensor([0.0386, 0.0924, 0.0519, 0.0447, 0.0575, 0.1004, 0.0479, 0.0597], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0260, 0.0375, 0.0330, 0.0272, 0.0306, 0.0311, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:39:22,749 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6125, 3.2945, 3.1378, 2.5934, 3.0504, 3.4172, 3.5932, 3.0243], device='cuda:1'), covar=tensor([0.0616, 0.1215, 0.0894, 0.1241, 0.1055, 0.0582, 0.1015, 0.1116], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0174, 0.0139, 0.0127, 0.0143, 0.0154, 0.0131, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:39:30,009 INFO [train.py:873] (1/4) Epoch 16, batch 300, loss[loss=0.1102, simple_loss=0.1452, pruned_loss=0.03755, over 14196.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1462, pruned_loss=0.03807, over 1614325.73 frames. ], batch size: 89, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:39:42,379 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9941, 1.9704, 1.4531, 1.5297, 1.9337, 1.9860, 1.9870, 1.9737], device='cuda:1'), covar=tensor([0.1493, 0.1031, 0.4332, 0.4405, 0.1654, 0.1717, 0.2169, 0.1592], device='cuda:1'), in_proj_covar=tensor([0.0385, 0.0265, 0.0443, 0.0567, 0.0341, 0.0441, 0.0389, 0.0386], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:40:28,277 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 1.970e+02 2.457e+02 3.061e+02 5.224e+02, threshold=4.913e+02, percent-clipped=0.0 2022-12-08 08:40:57,901 INFO [train.py:873] (1/4) Epoch 16, batch 400, loss[loss=0.1192, simple_loss=0.1219, pruned_loss=0.05822, over 2562.00 frames. ], tot_loss[loss=0.1109, simple_loss=0.1459, pruned_loss=0.03794, over 1734785.41 frames. ], batch size: 100, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:41:50,193 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113890.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:41:56,841 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.079e+02 2.042e+02 2.566e+02 3.444e+02 7.891e+02, threshold=5.133e+02, percent-clipped=8.0 2022-12-08 08:42:25,885 INFO [train.py:873] (1/4) Epoch 16, batch 500, loss[loss=0.1033, simple_loss=0.1487, pruned_loss=0.02893, over 13810.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1459, pruned_loss=0.03832, over 1818866.33 frames. ], batch size: 23, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:42:32,152 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113938.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:43:02,347 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:43:12,293 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7836, 4.5916, 4.1853, 4.3949, 4.5466, 4.7086, 4.7604, 4.7629], device='cuda:1'), covar=tensor([0.0692, 0.0415, 0.1972, 0.2378, 0.0651, 0.0685, 0.0684, 0.0717], device='cuda:1'), in_proj_covar=tensor([0.0392, 0.0269, 0.0449, 0.0575, 0.0342, 0.0448, 0.0396, 0.0392], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:43:24,305 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.124e+02 2.547e+02 2.966e+02 5.657e+02, threshold=5.095e+02, percent-clipped=2.0 2022-12-08 08:43:28,086 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:43:53,990 INFO [train.py:873] (1/4) Epoch 16, batch 600, loss[loss=0.09678, simple_loss=0.1392, pruned_loss=0.02716, over 14255.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.146, pruned_loss=0.03812, over 1909791.61 frames. ], batch size: 37, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:44:28,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6872, 4.8918, 5.4131, 5.7440, 5.2271, 4.5227, 5.6145, 4.5091], device='cuda:1'), covar=tensor([0.1046, 0.2367, 0.0850, 0.0902, 0.1616, 0.0768, 0.0948, 0.1238], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0267, 0.0195, 0.0190, 0.0183, 0.0154, 0.0280, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:44:36,040 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 08:44:52,421 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.140e+02 2.657e+02 3.260e+02 5.874e+02, threshold=5.314e+02, percent-clipped=5.0 2022-12-08 08:45:18,437 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:45:21,945 INFO [train.py:873] (1/4) Epoch 16, batch 700, loss[loss=0.1012, simple_loss=0.1442, pruned_loss=0.02907, over 14420.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1461, pruned_loss=0.03822, over 1977023.61 frames. ], batch size: 53, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:45:54,218 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4430, 1.4878, 1.5725, 1.4134, 1.2569, 1.2862, 1.0446, 1.0569], device='cuda:1'), covar=tensor([0.0180, 0.0251, 0.0171, 0.0204, 0.0207, 0.0337, 0.0263, 0.0360], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:46:12,504 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114188.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:46:20,991 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.016e+02 2.560e+02 3.238e+02 1.602e+03, threshold=5.120e+02, percent-clipped=5.0 2022-12-08 08:46:50,896 INFO [train.py:873] (1/4) Epoch 16, batch 800, loss[loss=0.09966, simple_loss=0.1409, pruned_loss=0.02919, over 13954.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.1463, pruned_loss=0.03793, over 1981962.03 frames. ], batch size: 23, lr: 4.98e-03, grad_scale: 8.0 2022-12-08 08:46:52,228 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-08 08:47:18,908 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2576, 2.0290, 2.1750, 2.3674, 2.0220, 1.9700, 2.2818, 2.2269], device='cuda:1'), covar=tensor([0.0285, 0.0551, 0.0279, 0.0263, 0.0468, 0.0665, 0.0333, 0.0318], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0258, 0.0373, 0.0329, 0.0271, 0.0305, 0.0310, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:47:21,703 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1827, 2.7648, 4.0731, 3.0028, 4.1160, 3.9292, 3.8496, 3.3503], device='cuda:1'), covar=tensor([0.0898, 0.2822, 0.0996, 0.1818, 0.0722, 0.0981, 0.1451, 0.2067], device='cuda:1'), in_proj_covar=tensor([0.0358, 0.0315, 0.0396, 0.0300, 0.0374, 0.0326, 0.0363, 0.0304], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:47:28,697 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:47:30,427 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1272, 3.9151, 3.6127, 3.7999, 4.0058, 4.0628, 4.1088, 4.1131], device='cuda:1'), covar=tensor([0.0915, 0.0561, 0.2349, 0.2719, 0.0808, 0.0964, 0.1008, 0.0962], device='cuda:1'), in_proj_covar=tensor([0.0391, 0.0268, 0.0451, 0.0572, 0.0344, 0.0448, 0.0394, 0.0393], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:47:39,651 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0282, 2.2094, 2.2448, 2.3231, 1.9855, 2.3296, 2.1350, 1.3161], device='cuda:1'), covar=tensor([0.0818, 0.0816, 0.0888, 0.0549, 0.1019, 0.0577, 0.1090, 0.2004], device='cuda:1'), in_proj_covar=tensor([0.0143, 0.0089, 0.0070, 0.0073, 0.0100, 0.0088, 0.0102, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:47:50,274 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.518e+01 2.109e+02 2.499e+02 3.135e+02 7.663e+02, threshold=4.999e+02, percent-clipped=3.0 2022-12-08 08:47:53,309 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9841, 2.5340, 3.9528, 4.0991, 3.8977, 2.3463, 4.0425, 3.1365], device='cuda:1'), covar=tensor([0.0453, 0.1112, 0.0962, 0.0502, 0.0511, 0.1810, 0.0407, 0.0939], device='cuda:1'), in_proj_covar=tensor([0.0290, 0.0258, 0.0372, 0.0328, 0.0271, 0.0305, 0.0309, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:47:54,012 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114302.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:48:09,876 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:48:18,623 INFO [train.py:873] (1/4) Epoch 16, batch 900, loss[loss=0.1398, simple_loss=0.159, pruned_loss=0.06032, over 7794.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1469, pruned_loss=0.03873, over 1931871.47 frames. ], batch size: 100, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:48:35,141 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:49:17,095 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.233e+02 2.852e+02 3.607e+02 7.620e+02, threshold=5.705e+02, percent-clipped=7.0 2022-12-08 08:49:44,141 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1141, 1.1337, 1.0115, 1.1349, 1.2792, 0.7983, 1.0226, 1.1362], device='cuda:1'), covar=tensor([0.0524, 0.0660, 0.0553, 0.0709, 0.0461, 0.0819, 0.0962, 0.0739], device='cuda:1'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0036], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 08:49:45,661 INFO [train.py:873] (1/4) Epoch 16, batch 1000, loss[loss=0.1239, simple_loss=0.1538, pruned_loss=0.04696, over 14330.00 frames. ], tot_loss[loss=0.1126, simple_loss=0.147, pruned_loss=0.03915, over 1935132.86 frames. ], batch size: 55, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:49:55,269 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8739, 2.0634, 2.8137, 2.8890, 2.8339, 2.2012, 2.8756, 2.3040], device='cuda:1'), covar=tensor([0.0411, 0.1118, 0.0623, 0.0474, 0.0572, 0.1399, 0.0446, 0.0825], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0260, 0.0375, 0.0330, 0.0274, 0.0307, 0.0312, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 08:50:01,092 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7948, 2.0435, 2.7453, 2.1582, 2.7509, 2.6453, 2.4815, 2.4037], device='cuda:1'), covar=tensor([0.0787, 0.3421, 0.1026, 0.1796, 0.0605, 0.1154, 0.0851, 0.1653], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0314, 0.0394, 0.0299, 0.0371, 0.0325, 0.0363, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:50:30,802 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:50:45,034 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.045e+02 2.472e+02 3.169e+02 1.892e+03, threshold=4.944e+02, percent-clipped=6.0 2022-12-08 08:50:46,113 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4007, 1.0020, 1.2141, 0.8856, 1.1313, 1.3671, 1.0773, 1.0863], device='cuda:1'), covar=tensor([0.0361, 0.0871, 0.0607, 0.0418, 0.0882, 0.0772, 0.0445, 0.1086], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0173, 0.0138, 0.0126, 0.0141, 0.0153, 0.0131, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:50:48,537 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2022-12-08 08:51:13,048 INFO [train.py:873] (1/4) Epoch 16, batch 1100, loss[loss=0.09794, simple_loss=0.1413, pruned_loss=0.0273, over 14152.00 frames. ], tot_loss[loss=0.1116, simple_loss=0.1463, pruned_loss=0.03838, over 1980875.11 frames. ], batch size: 84, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:51:35,955 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.41 vs. limit=2.0 2022-12-08 08:51:45,223 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7067, 1.9332, 2.6419, 2.1497, 2.6779, 2.5206, 2.3914, 2.3110], device='cuda:1'), covar=tensor([0.0849, 0.2856, 0.1064, 0.1657, 0.0666, 0.1425, 0.1071, 0.1445], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0315, 0.0395, 0.0299, 0.0372, 0.0325, 0.0363, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:52:01,162 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1323, 2.1637, 2.4729, 1.6000, 1.7929, 2.2182, 1.5017, 2.2667], device='cuda:1'), covar=tensor([0.1199, 0.1571, 0.0769, 0.2522, 0.2413, 0.1209, 0.3107, 0.0930], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0093, 0.0099, 0.0116, 0.0088, 0.0120, 0.0091], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 08:52:12,292 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.285e+02 2.811e+02 4.159e+02 1.083e+03, threshold=5.623e+02, percent-clipped=11.0 2022-12-08 08:52:34,288 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 08:52:41,092 INFO [train.py:873] (1/4) Epoch 16, batch 1200, loss[loss=0.1591, simple_loss=0.1386, pruned_loss=0.08977, over 1237.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1461, pruned_loss=0.03817, over 1990706.89 frames. ], batch size: 100, lr: 4.98e-03, grad_scale: 8.0 2022-12-08 08:52:42,995 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0137, 3.3262, 3.0752, 3.3324, 2.5551, 3.3242, 3.0948, 1.6273], device='cuda:1'), covar=tensor([0.1503, 0.0776, 0.1060, 0.0463, 0.0961, 0.0626, 0.1025, 0.2185], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0087, 0.0068, 0.0072, 0.0098, 0.0086, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:53:41,425 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 1.943e+02 2.456e+02 3.315e+02 7.474e+02, threshold=4.912e+02, percent-clipped=2.0 2022-12-08 08:53:56,734 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114716.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:54:09,740 INFO [train.py:873] (1/4) Epoch 16, batch 1300, loss[loss=0.1026, simple_loss=0.1465, pruned_loss=0.02938, over 14306.00 frames. ], tot_loss[loss=0.1105, simple_loss=0.1457, pruned_loss=0.03766, over 1970245.37 frames. ], batch size: 39, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:54:50,195 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:54:56,095 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114783.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:55:09,771 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.098e+02 2.728e+02 3.262e+02 8.407e+02, threshold=5.455e+02, percent-clipped=4.0 2022-12-08 08:55:38,356 INFO [train.py:873] (1/4) Epoch 16, batch 1400, loss[loss=0.1539, simple_loss=0.1726, pruned_loss=0.0676, over 7746.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1456, pruned_loss=0.03785, over 1986229.54 frames. ], batch size: 100, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:55:38,423 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:56:14,186 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114871.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:56:38,331 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 2.101e+02 2.672e+02 3.477e+02 6.209e+02, threshold=5.345e+02, percent-clipped=3.0 2022-12-08 08:57:06,855 INFO [train.py:873] (1/4) Epoch 16, batch 1500, loss[loss=0.1173, simple_loss=0.1358, pruned_loss=0.04944, over 3852.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1457, pruned_loss=0.03838, over 1983220.73 frames. ], batch size: 100, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:57:07,957 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114932.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:57:41,463 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6083, 3.7729, 3.9159, 3.6178, 3.7873, 3.7833, 1.5103, 3.5928], device='cuda:1'), covar=tensor([0.0385, 0.0370, 0.0323, 0.0415, 0.0338, 0.0454, 0.3085, 0.0309], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0172, 0.0143, 0.0144, 0.0203, 0.0140, 0.0156, 0.0191], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 08:57:45,728 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0677, 1.5055, 3.1893, 1.5069, 3.1531, 3.2026, 2.3425, 3.4192], device='cuda:1'), covar=tensor([0.0292, 0.2963, 0.0463, 0.2285, 0.0763, 0.0476, 0.0977, 0.0246], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0159, 0.0168, 0.0167, 0.0178, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 08:58:07,016 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.137e+02 2.682e+02 3.588e+02 7.078e+02, threshold=5.364e+02, percent-clipped=5.0 2022-12-08 08:58:39,010 INFO [train.py:873] (1/4) Epoch 16, batch 1600, loss[loss=0.1072, simple_loss=0.1503, pruned_loss=0.03206, over 14312.00 frames. ], tot_loss[loss=0.11, simple_loss=0.1455, pruned_loss=0.03723, over 2036345.92 frames. ], batch size: 46, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:58:39,206 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4227, 3.6996, 3.3751, 3.6643, 2.6667, 3.6837, 3.4724, 2.0541], device='cuda:1'), covar=tensor([0.1373, 0.0933, 0.1035, 0.0603, 0.0937, 0.0659, 0.0992, 0.1941], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0087, 0.0069, 0.0072, 0.0098, 0.0086, 0.0100, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 08:59:15,334 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:59:16,492 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3998, 4.9250, 4.8464, 5.3533, 4.9438, 4.5160, 5.3215, 4.4891], device='cuda:1'), covar=tensor([0.0338, 0.1122, 0.0364, 0.0433, 0.0921, 0.0546, 0.0544, 0.0482], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0277, 0.0201, 0.0196, 0.0188, 0.0161, 0.0288, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 08:59:17,792 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.69 vs. limit=5.0 2022-12-08 08:59:19,738 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0263, 2.0283, 2.0911, 2.0991, 2.0268, 1.7132, 1.3687, 1.8757], device='cuda:1'), covar=tensor([0.0697, 0.0565, 0.0523, 0.0360, 0.0478, 0.1388, 0.2239, 0.0516], device='cuda:1'), in_proj_covar=tensor([0.0168, 0.0170, 0.0142, 0.0143, 0.0201, 0.0139, 0.0154, 0.0189], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 08:59:27,760 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115086.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:59:39,793 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.275e+02 2.932e+02 3.523e+02 7.695e+02, threshold=5.865e+02, percent-clipped=6.0 2022-12-08 08:59:40,116 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5074, 2.2673, 3.3495, 3.5590, 3.4023, 2.2403, 3.3650, 2.6703], device='cuda:1'), covar=tensor([0.0438, 0.1159, 0.0825, 0.0455, 0.0513, 0.1658, 0.0472, 0.0986], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0260, 0.0380, 0.0331, 0.0273, 0.0308, 0.0313, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 09:00:07,459 INFO [train.py:873] (1/4) Epoch 16, batch 1700, loss[loss=0.1298, simple_loss=0.1584, pruned_loss=0.05057, over 10327.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.146, pruned_loss=0.03829, over 1983805.88 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:00:17,270 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3741, 2.2951, 3.3705, 3.4641, 3.3556, 2.2414, 3.3174, 2.5951], device='cuda:1'), covar=tensor([0.0467, 0.1168, 0.0738, 0.0541, 0.0540, 0.1616, 0.0480, 0.1057], device='cuda:1'), in_proj_covar=tensor([0.0291, 0.0258, 0.0377, 0.0328, 0.0271, 0.0306, 0.0310, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 09:00:21,736 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:00:21,753 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8234, 1.8565, 1.6343, 1.9248, 1.8359, 1.8657, 1.7305, 1.7295], device='cuda:1'), covar=tensor([0.1020, 0.0818, 0.1882, 0.0789, 0.1024, 0.0586, 0.1345, 0.1075], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0292, 0.0260, 0.0280, 0.0321, 0.0299, 0.0254, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:01:09,161 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.404e+01 2.088e+02 2.519e+02 3.191e+02 6.447e+02, threshold=5.039e+02, percent-clipped=1.0 2022-12-08 09:01:33,714 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115227.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:01:36,926 INFO [train.py:873] (1/4) Epoch 16, batch 1800, loss[loss=0.1541, simple_loss=0.1735, pruned_loss=0.06739, over 10366.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1459, pruned_loss=0.0384, over 1959737.11 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:02:24,783 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3677, 2.0467, 2.3506, 2.4332, 2.2604, 1.9297, 2.3793, 2.1893], device='cuda:1'), covar=tensor([0.0413, 0.0959, 0.0510, 0.0390, 0.0616, 0.1140, 0.0492, 0.0639], device='cuda:1'), in_proj_covar=tensor([0.0296, 0.0262, 0.0382, 0.0332, 0.0275, 0.0311, 0.0315, 0.0285], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 09:02:32,637 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2489, 2.0968, 2.1618, 2.2397, 2.1186, 2.1533, 2.3324, 1.9736], device='cuda:1'), covar=tensor([0.0723, 0.1083, 0.0708, 0.0771, 0.1044, 0.0628, 0.0803, 0.0632], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0274, 0.0198, 0.0194, 0.0186, 0.0158, 0.0284, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:02:38,559 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.091e+02 1.953e+02 2.460e+02 2.998e+02 4.316e+02, threshold=4.920e+02, percent-clipped=0.0 2022-12-08 09:03:06,676 INFO [train.py:873] (1/4) Epoch 16, batch 1900, loss[loss=0.09639, simple_loss=0.1457, pruned_loss=0.02352, over 13968.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.1459, pruned_loss=0.03817, over 1979394.47 frames. ], batch size: 23, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:03:12,568 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 09:03:41,648 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115370.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:03:42,498 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:03:43,285 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115372.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:03:50,896 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1358, 3.9195, 3.8737, 4.2249, 3.7471, 3.6024, 4.2286, 4.0110], device='cuda:1'), covar=tensor([0.0723, 0.0995, 0.0820, 0.0649, 0.0871, 0.0661, 0.0645, 0.0797], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0142, 0.0145, 0.0159, 0.0147, 0.0124, 0.0167, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 09:04:08,304 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.103e+02 2.568e+02 3.222e+02 1.079e+03, threshold=5.135e+02, percent-clipped=6.0 2022-12-08 09:04:14,629 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-08 09:04:26,553 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:36,093 INFO [train.py:873] (1/4) Epoch 16, batch 2000, loss[loss=0.1626, simple_loss=0.1593, pruned_loss=0.08294, over 1286.00 frames. ], tot_loss[loss=0.1116, simple_loss=0.1458, pruned_loss=0.03865, over 1943421.09 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 8.0 2022-12-08 09:04:36,241 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115431.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:37,107 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:40,858 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4740, 2.3549, 3.3640, 2.6263, 3.4235, 3.2829, 3.1592, 2.8651], device='cuda:1'), covar=tensor([0.1011, 0.2815, 0.1151, 0.1942, 0.0902, 0.1103, 0.1321, 0.1817], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0313, 0.0398, 0.0301, 0.0372, 0.0326, 0.0365, 0.0301], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:04:45,847 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115442.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:53,601 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7462, 2.6394, 2.0647, 2.7810, 2.6483, 2.6821, 2.3606, 2.1638], device='cuda:1'), covar=tensor([0.1176, 0.1279, 0.2736, 0.0901, 0.1250, 0.1091, 0.1466, 0.2361], device='cuda:1'), in_proj_covar=tensor([0.0280, 0.0292, 0.0259, 0.0280, 0.0320, 0.0299, 0.0253, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:05:14,060 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4583, 2.7300, 4.2253, 3.2219, 4.2598, 4.0740, 4.0270, 3.5998], device='cuda:1'), covar=tensor([0.0766, 0.2772, 0.0826, 0.1524, 0.0633, 0.0830, 0.1207, 0.1659], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0313, 0.0397, 0.0301, 0.0371, 0.0325, 0.0364, 0.0300], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:05:36,028 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.145e+01 1.975e+02 2.632e+02 3.322e+02 6.213e+02, threshold=5.264e+02, percent-clipped=1.0 2022-12-08 09:05:59,700 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115527.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:06:03,272 INFO [train.py:873] (1/4) Epoch 16, batch 2100, loss[loss=0.108, simple_loss=0.141, pruned_loss=0.03753, over 13891.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1453, pruned_loss=0.03798, over 2009017.87 frames. ], batch size: 23, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:06:29,912 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8645, 2.5112, 4.8619, 3.2667, 4.5895, 2.2207, 3.6387, 4.6603], device='cuda:1'), covar=tensor([0.0385, 0.3503, 0.0300, 0.6013, 0.0486, 0.3060, 0.1117, 0.0290], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0203, 0.0214, 0.0274, 0.0233, 0.0203, 0.0204, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:06:42,184 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:07:05,100 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 1.971e+02 2.573e+02 2.986e+02 1.089e+03, threshold=5.147e+02, percent-clipped=6.0 2022-12-08 09:07:31,194 INFO [train.py:873] (1/4) Epoch 16, batch 2200, loss[loss=0.1075, simple_loss=0.1411, pruned_loss=0.03693, over 11133.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1458, pruned_loss=0.03831, over 2039569.50 frames. ], batch size: 100, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:08:00,339 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4051, 1.5894, 4.2157, 1.7444, 4.1872, 4.4643, 3.8809, 4.8198], device='cuda:1'), covar=tensor([0.0220, 0.3087, 0.0466, 0.2346, 0.0393, 0.0365, 0.0403, 0.0169], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0160, 0.0168, 0.0166, 0.0178, 0.0134, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:08:32,753 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.098e+02 2.631e+02 3.239e+02 6.147e+02, threshold=5.263e+02, percent-clipped=2.0 2022-12-08 09:08:54,845 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115726.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:08:55,719 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115727.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:08:58,973 INFO [train.py:873] (1/4) Epoch 16, batch 2300, loss[loss=0.1216, simple_loss=0.135, pruned_loss=0.05413, over 2573.00 frames. ], tot_loss[loss=0.11, simple_loss=0.1451, pruned_loss=0.03748, over 1990901.59 frames. ], batch size: 100, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:09:08,632 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115742.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:09:12,341 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 09:09:28,145 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 09:09:50,752 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115790.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:10:00,477 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.978e+01 2.138e+02 2.481e+02 3.114e+02 7.710e+02, threshold=4.963e+02, percent-clipped=3.0 2022-12-08 09:10:26,759 INFO [train.py:873] (1/4) Epoch 16, batch 2400, loss[loss=0.1107, simple_loss=0.1455, pruned_loss=0.03799, over 14393.00 frames. ], tot_loss[loss=0.1096, simple_loss=0.1451, pruned_loss=0.03708, over 2038289.48 frames. ], batch size: 53, lr: 4.95e-03, grad_scale: 8.0 2022-12-08 09:10:28,899 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2022-12-08 09:11:19,651 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115892.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:11:27,447 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.676e+01 2.182e+02 2.667e+02 3.375e+02 7.141e+02, threshold=5.334e+02, percent-clipped=4.0 2022-12-08 09:11:54,002 INFO [train.py:873] (1/4) Epoch 16, batch 2500, loss[loss=0.126, simple_loss=0.1316, pruned_loss=0.06021, over 2657.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.1452, pruned_loss=0.03748, over 1946159.97 frames. ], batch size: 100, lr: 4.95e-03, grad_scale: 8.0 2022-12-08 09:12:13,578 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115953.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:12:56,497 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.068e+02 2.436e+02 3.138e+02 6.345e+02, threshold=4.873e+02, percent-clipped=1.0 2022-12-08 09:13:03,554 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8162, 2.0006, 2.1360, 2.1829, 1.9881, 2.1586, 1.9043, 1.4432], device='cuda:1'), covar=tensor([0.1024, 0.1163, 0.0678, 0.0653, 0.1128, 0.0790, 0.1398, 0.2069], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0088, 0.0069, 0.0072, 0.0098, 0.0088, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:13:07,091 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116014.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:11,577 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6598, 1.7819, 1.7361, 1.7051, 1.7279, 1.2528, 1.4829, 1.6390], device='cuda:1'), covar=tensor([0.1022, 0.0693, 0.0765, 0.0698, 0.0641, 0.0907, 0.0703, 0.0647], device='cuda:1'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0031, 0.0032, 0.0045, 0.0034, 0.0036], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 09:13:17,565 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:18,388 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:21,682 INFO [train.py:873] (1/4) Epoch 16, batch 2600, loss[loss=0.1356, simple_loss=0.1631, pruned_loss=0.054, over 14230.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1454, pruned_loss=0.03785, over 1989909.98 frames. ], batch size: 69, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:13:21,770 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2620, 3.9884, 3.8488, 4.2834, 3.9947, 3.8662, 4.3080, 3.6647], device='cuda:1'), covar=tensor([0.0432, 0.0906, 0.0446, 0.0417, 0.0700, 0.1079, 0.0489, 0.0473], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0268, 0.0195, 0.0192, 0.0182, 0.0155, 0.0279, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:13:34,588 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-08 09:13:59,622 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116074.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:00,557 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:00,708 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116075.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:14:23,795 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.134e+02 2.737e+02 3.520e+02 9.731e+02, threshold=5.473e+02, percent-clipped=5.0 2022-12-08 09:14:26,820 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116105.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:36,365 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:49,180 INFO [train.py:873] (1/4) Epoch 16, batch 2700, loss[loss=0.1127, simple_loss=0.1389, pruned_loss=0.04325, over 5035.00 frames. ], tot_loss[loss=0.1108, simple_loss=0.1454, pruned_loss=0.03807, over 1981288.72 frames. ], batch size: 100, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:15:20,040 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116166.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:15:29,430 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116177.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:15:51,299 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.057e+02 2.028e+02 2.481e+02 3.066e+02 5.287e+02, threshold=4.961e+02, percent-clipped=0.0 2022-12-08 09:16:16,288 INFO [train.py:873] (1/4) Epoch 16, batch 2800, loss[loss=0.1261, simple_loss=0.1381, pruned_loss=0.05705, over 2640.00 frames. ], tot_loss[loss=0.112, simple_loss=0.1461, pruned_loss=0.03894, over 1916558.24 frames. ], batch size: 100, lr: 4.94e-03, grad_scale: 8.0 2022-12-08 09:16:27,484 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3553, 0.9812, 1.1813, 0.8350, 1.1588, 1.3840, 1.0364, 1.1396], device='cuda:1'), covar=tensor([0.0504, 0.1034, 0.0889, 0.0631, 0.0853, 0.0833, 0.0617, 0.1161], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0175, 0.0139, 0.0127, 0.0142, 0.0153, 0.0133, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:16:31,690 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116248.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 09:16:45,661 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8619, 1.8957, 2.0120, 1.9503, 1.8819, 1.6818, 1.6032, 1.2348], device='cuda:1'), covar=tensor([0.0234, 0.0335, 0.0239, 0.0231, 0.0336, 0.0309, 0.0282, 0.0542], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 09:17:19,503 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.003e+02 2.440e+02 3.142e+02 5.921e+02, threshold=4.880e+02, percent-clipped=3.0 2022-12-08 09:17:37,644 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116323.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:17:38,961 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 09:17:44,502 INFO [train.py:873] (1/4) Epoch 16, batch 2900, loss[loss=0.1042, simple_loss=0.1163, pruned_loss=0.04602, over 2593.00 frames. ], tot_loss[loss=0.111, simple_loss=0.1456, pruned_loss=0.03821, over 1911335.93 frames. ], batch size: 100, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:18:18,475 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116370.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:18:30,451 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:18:43,981 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9892, 2.1136, 2.9239, 2.2592, 2.9134, 2.7866, 2.7938, 2.4696], device='cuda:1'), covar=tensor([0.0679, 0.2545, 0.0898, 0.1863, 0.0819, 0.0896, 0.0962, 0.1634], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0312, 0.0393, 0.0299, 0.0370, 0.0323, 0.0362, 0.0300], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:18:47,320 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.101e+02 2.585e+02 3.248e+02 5.004e+02, threshold=5.170e+02, percent-clipped=1.0 2022-12-08 09:19:11,752 INFO [train.py:873] (1/4) Epoch 16, batch 3000, loss[loss=0.1292, simple_loss=0.1609, pruned_loss=0.04871, over 14219.00 frames. ], tot_loss[loss=0.11, simple_loss=0.145, pruned_loss=0.0375, over 1979437.24 frames. ], batch size: 89, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:19:11,752 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 09:19:20,203 INFO [train.py:905] (1/4) Epoch 16, validation: loss=0.1369, simple_loss=0.1741, pruned_loss=0.04986, over 857387.00 frames. 2022-12-08 09:19:20,204 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 09:19:33,002 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116446.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:19:45,765 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116461.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:19:55,320 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:20:22,409 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.145e+02 2.640e+02 3.351e+02 6.789e+02, threshold=5.280e+02, percent-clipped=3.0 2022-12-08 09:20:24,280 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:20:26,031 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116507.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:20:40,639 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1359, 1.2503, 1.3182, 1.0438, 0.8516, 1.1105, 0.8242, 1.2207], device='cuda:1'), covar=tensor([0.1626, 0.2764, 0.1113, 0.2108, 0.2955, 0.1236, 0.1601, 0.1159], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0093, 0.0098, 0.0114, 0.0088, 0.0118, 0.0092], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 09:20:46,688 INFO [train.py:873] (1/4) Epoch 16, batch 3100, loss[loss=0.1156, simple_loss=0.1537, pruned_loss=0.03873, over 14234.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1454, pruned_loss=0.03796, over 1925191.23 frames. ], batch size: 25, lr: 4.93e-03, grad_scale: 4.0 2022-12-08 09:21:01,804 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116548.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:21:04,525 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2316, 1.5995, 2.4569, 2.0532, 2.2622, 1.6479, 1.9982, 2.2318], device='cuda:1'), covar=tensor([0.2017, 0.3524, 0.0786, 0.2071, 0.1420, 0.2227, 0.0991, 0.0901], device='cuda:1'), in_proj_covar=tensor([0.0254, 0.0204, 0.0215, 0.0275, 0.0235, 0.0206, 0.0205, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 09:21:12,998 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2723, 1.4481, 3.4312, 1.6146, 3.1361, 3.4198, 2.4381, 3.6421], device='cuda:1'), covar=tensor([0.0279, 0.3095, 0.0405, 0.2326, 0.0927, 0.0436, 0.0997, 0.0224], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0160, 0.0169, 0.0166, 0.0179, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:21:17,530 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116566.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:21:35,747 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2954, 3.0840, 3.0618, 3.2851, 3.1518, 3.2839, 3.3528, 2.7865], device='cuda:1'), covar=tensor([0.0535, 0.0987, 0.0538, 0.0565, 0.0741, 0.0417, 0.0591, 0.0581], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0269, 0.0196, 0.0192, 0.0182, 0.0155, 0.0280, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:21:43,616 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116596.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:21:49,521 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.181e+02 2.695e+02 3.453e+02 7.270e+02, threshold=5.390e+02, percent-clipped=4.0 2022-12-08 09:22:02,718 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116618.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:22:13,975 INFO [train.py:873] (1/4) Epoch 16, batch 3200, loss[loss=0.1304, simple_loss=0.1625, pruned_loss=0.04918, over 14507.00 frames. ], tot_loss[loss=0.1117, simple_loss=0.1462, pruned_loss=0.03864, over 1926059.50 frames. ], batch size: 49, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:22:17,267 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-08 09:22:42,782 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8020, 1.2632, 1.6348, 1.1764, 1.5422, 1.8400, 1.4623, 1.5038], device='cuda:1'), covar=tensor([0.0540, 0.0695, 0.0635, 0.0830, 0.1173, 0.0729, 0.0675, 0.1430], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0174, 0.0139, 0.0126, 0.0142, 0.0153, 0.0133, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:22:48,032 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116670.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:22:51,538 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6000, 1.7198, 1.8155, 1.3235, 1.3095, 1.6080, 1.2297, 1.6590], device='cuda:1'), covar=tensor([0.1522, 0.2468, 0.0957, 0.2701, 0.3043, 0.1171, 0.2809, 0.1211], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0094, 0.0099, 0.0116, 0.0089, 0.0119, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 09:22:55,616 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:22:55,722 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:22:58,949 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1715, 3.4209, 3.2997, 3.2858, 2.5602, 3.4770, 3.3298, 1.8070], device='cuda:1'), covar=tensor([0.1650, 0.1687, 0.0997, 0.0801, 0.1030, 0.0467, 0.0998, 0.2272], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0088, 0.0069, 0.0072, 0.0098, 0.0087, 0.0100, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:23:16,519 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.227e+02 2.604e+02 3.188e+02 6.019e+02, threshold=5.207e+02, percent-clipped=1.0 2022-12-08 09:23:17,676 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:23:29,699 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:23:35,532 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 09:23:41,059 INFO [train.py:873] (1/4) Epoch 16, batch 3300, loss[loss=0.1063, simple_loss=0.1454, pruned_loss=0.03361, over 14420.00 frames. ], tot_loss[loss=0.1118, simple_loss=0.1462, pruned_loss=0.03867, over 1934493.09 frames. ], batch size: 53, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:24:07,166 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116761.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:10,619 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:16,809 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:43,064 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116802.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 09:24:43,136 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116802.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:43,678 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 1.987e+02 2.396e+02 2.955e+02 5.766e+02, threshold=4.792e+02, percent-clipped=2.0 2022-12-08 09:24:48,965 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116809.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:58,338 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:25:08,246 INFO [train.py:873] (1/4) Epoch 16, batch 3400, loss[loss=0.09784, simple_loss=0.1371, pruned_loss=0.0293, over 11214.00 frames. ], tot_loss[loss=0.112, simple_loss=0.1461, pruned_loss=0.03891, over 1921607.70 frames. ], batch size: 100, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:25:26,900 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 09:25:34,074 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116861.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:25:35,760 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9151, 1.5719, 3.9954, 1.9169, 3.8401, 4.0704, 3.1304, 4.2270], device='cuda:1'), covar=tensor([0.0296, 0.3441, 0.0428, 0.2420, 0.0544, 0.0420, 0.0733, 0.0277], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0155, 0.0159, 0.0168, 0.0166, 0.0179, 0.0134, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:25:35,838 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116863.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:25:40,957 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 09:26:06,918 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.45 vs. limit=2.0 2022-12-08 09:26:10,487 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.102e+02 2.045e+02 2.562e+02 3.256e+02 5.626e+02, threshold=5.125e+02, percent-clipped=2.0 2022-12-08 09:26:35,745 INFO [train.py:873] (1/4) Epoch 16, batch 3500, loss[loss=0.115, simple_loss=0.15, pruned_loss=0.03997, over 14221.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1457, pruned_loss=0.03832, over 1994661.73 frames. ], batch size: 60, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:26:36,653 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9826, 1.9475, 2.0601, 2.0713, 2.0063, 1.6666, 1.3101, 1.8090], device='cuda:1'), covar=tensor([0.0777, 0.0686, 0.0615, 0.0436, 0.0550, 0.1576, 0.2413, 0.0564], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0172, 0.0145, 0.0146, 0.0204, 0.0141, 0.0156, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 09:26:39,816 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1242, 2.1588, 2.3637, 1.6217, 1.6533, 2.1684, 1.2993, 2.1931], device='cuda:1'), covar=tensor([0.1067, 0.1591, 0.0847, 0.2414, 0.2858, 0.1003, 0.3781, 0.0949], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0102, 0.0095, 0.0100, 0.0116, 0.0089, 0.0120, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 09:27:12,581 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:27:16,931 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116979.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:27:18,679 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5927, 3.3035, 2.6552, 3.6449, 3.5242, 3.5447, 3.2338, 2.5609], device='cuda:1'), covar=tensor([0.0725, 0.1217, 0.2827, 0.0677, 0.0829, 0.1192, 0.1085, 0.3034], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0293, 0.0260, 0.0283, 0.0321, 0.0303, 0.0256, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:27:38,633 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.091e+02 2.141e+02 2.605e+02 3.184e+02 8.184e+02, threshold=5.210e+02, percent-clipped=2.0 2022-12-08 09:27:58,636 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:28:02,010 INFO [train.py:873] (1/4) Epoch 16, batch 3600, loss[loss=0.1171, simple_loss=0.1494, pruned_loss=0.0424, over 14217.00 frames. ], tot_loss[loss=0.1104, simple_loss=0.1452, pruned_loss=0.03783, over 1958034.59 frames. ], batch size: 94, lr: 4.92e-03, grad_scale: 8.0 2022-12-08 09:28:27,265 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117060.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:03,712 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117102.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:06,091 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.154e+02 2.650e+02 3.296e+02 7.753e+02, threshold=5.300e+02, percent-clipped=2.0 2022-12-08 09:29:21,954 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6073, 2.4711, 2.1870, 2.3517, 2.5525, 2.5731, 2.5899, 2.5804], device='cuda:1'), covar=tensor([0.1024, 0.0866, 0.2753, 0.2589, 0.1111, 0.1120, 0.1356, 0.0956], device='cuda:1'), in_proj_covar=tensor([0.0386, 0.0269, 0.0449, 0.0565, 0.0348, 0.0446, 0.0390, 0.0391], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:29:28,980 INFO [train.py:873] (1/4) Epoch 16, batch 3700, loss[loss=0.1318, simple_loss=0.1589, pruned_loss=0.05238, over 13539.00 frames. ], tot_loss[loss=0.1112, simple_loss=0.1459, pruned_loss=0.03829, over 1973093.46 frames. ], batch size: 100, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:29:30,776 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0588, 3.8846, 3.5587, 3.7059, 3.9857, 4.0181, 4.0984, 4.0839], device='cuda:1'), covar=tensor([0.0889, 0.0570, 0.2173, 0.2659, 0.0775, 0.0891, 0.0886, 0.0792], device='cuda:1'), in_proj_covar=tensor([0.0387, 0.0269, 0.0449, 0.0566, 0.0348, 0.0446, 0.0389, 0.0392], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:29:32,526 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9459, 1.6139, 1.8856, 1.4151, 1.7249, 2.0443, 1.8248, 1.7316], device='cuda:1'), covar=tensor([0.0861, 0.0722, 0.0808, 0.1263, 0.1695, 0.0828, 0.0874, 0.1857], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0174, 0.0139, 0.0125, 0.0141, 0.0152, 0.0132, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:29:45,674 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117150.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:52,495 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:55,082 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117161.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:10,628 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7842, 1.6166, 3.7641, 1.8073, 3.6695, 3.9444, 2.7010, 4.1448], device='cuda:1'), covar=tensor([0.0238, 0.2887, 0.0418, 0.2082, 0.0521, 0.0346, 0.0881, 0.0184], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0156, 0.0160, 0.0169, 0.0168, 0.0180, 0.0134, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:30:17,108 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2749, 2.9306, 5.2783, 3.6241, 4.8729, 2.4875, 3.7862, 4.9389], device='cuda:1'), covar=tensor([0.0380, 0.3120, 0.0268, 0.5202, 0.0463, 0.3151, 0.1221, 0.0353], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0201, 0.0213, 0.0270, 0.0231, 0.0204, 0.0201, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:30:27,523 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7357, 5.2452, 5.1803, 5.6718, 5.3531, 4.6482, 5.6574, 4.5008], device='cuda:1'), covar=tensor([0.0315, 0.0960, 0.0356, 0.0339, 0.0641, 0.0411, 0.0424, 0.0527], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0270, 0.0198, 0.0194, 0.0183, 0.0157, 0.0282, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:30:34,102 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.079e+02 2.495e+02 3.268e+02 6.334e+02, threshold=4.990e+02, percent-clipped=4.0 2022-12-08 09:30:37,997 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117209.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:48,169 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:57,050 INFO [train.py:873] (1/4) Epoch 16, batch 3800, loss[loss=0.0922, simple_loss=0.1347, pruned_loss=0.02485, over 14097.00 frames. ], tot_loss[loss=0.1099, simple_loss=0.145, pruned_loss=0.03742, over 1990033.92 frames. ], batch size: 29, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:31:18,267 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.42 vs. limit=2.0 2022-12-08 09:31:34,985 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117274.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:31:41,822 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117282.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:31:48,535 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3428, 3.1135, 2.4054, 3.4425, 3.3132, 3.3737, 2.9537, 2.4170], device='cuda:1'), covar=tensor([0.0975, 0.1331, 0.3144, 0.0705, 0.0985, 0.0856, 0.1317, 0.2866], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0292, 0.0260, 0.0283, 0.0320, 0.0300, 0.0255, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:32:02,068 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.002e+02 2.660e+02 3.371e+02 8.415e+02, threshold=5.321e+02, percent-clipped=3.0 2022-12-08 09:32:07,241 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 09:32:17,226 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117322.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:32:23,686 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:32:25,197 INFO [train.py:873] (1/4) Epoch 16, batch 3900, loss[loss=0.125, simple_loss=0.1496, pruned_loss=0.05019, over 7788.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1439, pruned_loss=0.03656, over 1966998.11 frames. ], batch size: 100, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:32:27,014 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8625, 2.7771, 2.0685, 2.8676, 2.7302, 2.7737, 2.4392, 2.2267], device='cuda:1'), covar=tensor([0.0861, 0.1277, 0.3563, 0.0927, 0.1331, 0.0896, 0.1594, 0.2798], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0293, 0.0261, 0.0284, 0.0321, 0.0301, 0.0256, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:32:51,273 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117360.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:17,690 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:18,728 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 09:33:24,023 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8181, 1.5126, 2.0424, 1.6267, 1.8742, 1.3960, 1.6892, 1.9210], device='cuda:1'), covar=tensor([0.3227, 0.2923, 0.0707, 0.1729, 0.1701, 0.1688, 0.1209, 0.1087], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0203, 0.0215, 0.0273, 0.0232, 0.0206, 0.0203, 0.0218], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 09:33:30,799 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 1.882e+02 2.302e+02 2.791e+02 6.246e+02, threshold=4.605e+02, percent-clipped=1.0 2022-12-08 09:33:33,442 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:53,716 INFO [train.py:873] (1/4) Epoch 16, batch 4000, loss[loss=0.08291, simple_loss=0.1283, pruned_loss=0.01876, over 13929.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1437, pruned_loss=0.03635, over 1982643.05 frames. ], batch size: 26, lr: 4.92e-03, grad_scale: 8.0 2022-12-08 09:34:17,482 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117458.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:34:58,373 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 09:34:58,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.165e+02 2.594e+02 3.323e+02 1.177e+03, threshold=5.188e+02, percent-clipped=6.0 2022-12-08 09:35:00,063 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:35:22,146 INFO [train.py:873] (1/4) Epoch 16, batch 4100, loss[loss=0.1173, simple_loss=0.1443, pruned_loss=0.04517, over 9470.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1442, pruned_loss=0.0369, over 1959720.72 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:36:02,202 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:36:27,068 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.013e+02 2.651e+02 3.289e+02 6.397e+02, threshold=5.302e+02, percent-clipped=7.0 2022-12-08 09:36:49,546 INFO [train.py:873] (1/4) Epoch 16, batch 4200, loss[loss=0.1603, simple_loss=0.1507, pruned_loss=0.08493, over 1268.00 frames. ], tot_loss[loss=0.1095, simple_loss=0.1446, pruned_loss=0.03718, over 1925903.97 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:37:37,410 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:37:45,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7853, 1.7489, 4.6318, 2.3476, 4.4769, 4.8106, 4.4073, 5.2069], device='cuda:1'), covar=tensor([0.0203, 0.3019, 0.0299, 0.1935, 0.0279, 0.0351, 0.0279, 0.0133], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0155, 0.0160, 0.0169, 0.0167, 0.0180, 0.0133, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:37:54,402 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.184e+02 2.504e+02 3.069e+02 5.464e+02, threshold=5.008e+02, percent-clipped=1.0 2022-12-08 09:37:57,164 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7513, 4.4036, 4.2560, 4.7393, 4.4120, 4.1476, 4.7721, 3.9776], device='cuda:1'), covar=tensor([0.0401, 0.0940, 0.0404, 0.0411, 0.0796, 0.0705, 0.0468, 0.0519], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0274, 0.0200, 0.0194, 0.0185, 0.0158, 0.0283, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:38:14,120 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0983, 2.5593, 3.5219, 2.3846, 2.0975, 2.9988, 1.7233, 2.8370], device='cuda:1'), covar=tensor([0.0612, 0.1201, 0.0458, 0.1771, 0.2078, 0.0669, 0.3036, 0.1158], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0102, 0.0095, 0.0101, 0.0117, 0.0089, 0.0121, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 09:38:17,333 INFO [train.py:873] (1/4) Epoch 16, batch 4300, loss[loss=0.109, simple_loss=0.1463, pruned_loss=0.03584, over 14253.00 frames. ], tot_loss[loss=0.1098, simple_loss=0.145, pruned_loss=0.03727, over 1984072.02 frames. ], batch size: 63, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:38:29,009 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4102, 2.4548, 2.5338, 2.4794, 2.5199, 2.1246, 1.4148, 2.2652], device='cuda:1'), covar=tensor([0.0589, 0.0525, 0.0442, 0.0440, 0.0417, 0.1324, 0.2573, 0.0446], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0172, 0.0146, 0.0145, 0.0204, 0.0140, 0.0156, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 09:38:31,014 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1945, 3.9824, 3.6748, 3.8423, 4.0717, 4.1316, 4.1697, 4.1844], device='cuda:1'), covar=tensor([0.0769, 0.0502, 0.2075, 0.2488, 0.0699, 0.0847, 0.0896, 0.0800], device='cuda:1'), in_proj_covar=tensor([0.0389, 0.0268, 0.0449, 0.0563, 0.0347, 0.0444, 0.0385, 0.0392], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:39:23,470 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.083e+02 2.539e+02 3.211e+02 7.278e+02, threshold=5.077e+02, percent-clipped=7.0 2022-12-08 09:39:31,549 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8669, 1.8844, 1.6363, 1.9155, 1.7632, 1.8210, 1.8385, 1.6770], device='cuda:1'), covar=tensor([0.1226, 0.0850, 0.1909, 0.0998, 0.1132, 0.0713, 0.1459, 0.0884], device='cuda:1'), in_proj_covar=tensor([0.0286, 0.0293, 0.0262, 0.0285, 0.0324, 0.0302, 0.0255, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:39:46,097 INFO [train.py:873] (1/4) Epoch 16, batch 4400, loss[loss=0.1264, simple_loss=0.145, pruned_loss=0.05392, over 3844.00 frames. ], tot_loss[loss=0.1097, simple_loss=0.1452, pruned_loss=0.03714, over 2020915.86 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:40:26,872 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117877.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:40:45,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2374, 2.5499, 2.5818, 2.6191, 2.1097, 2.5962, 2.4013, 1.3947], device='cuda:1'), covar=tensor([0.1163, 0.0814, 0.0574, 0.0506, 0.0941, 0.0676, 0.0941, 0.2074], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0088, 0.0069, 0.0073, 0.0098, 0.0088, 0.0101, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 09:40:47,199 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 09:40:50,806 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.093e+02 2.556e+02 3.184e+02 6.806e+02, threshold=5.111e+02, percent-clipped=1.0 2022-12-08 09:41:09,096 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:41:14,049 INFO [train.py:873] (1/4) Epoch 16, batch 4500, loss[loss=0.1138, simple_loss=0.1187, pruned_loss=0.05444, over 1303.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.145, pruned_loss=0.03635, over 2059286.80 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:41:15,957 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4359, 1.5330, 4.3982, 2.1320, 4.2592, 4.5205, 3.8565, 4.8070], device='cuda:1'), covar=tensor([0.0223, 0.3260, 0.0368, 0.2099, 0.0325, 0.0393, 0.0473, 0.0181], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0155, 0.0159, 0.0169, 0.0167, 0.0179, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:42:01,168 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:42:18,930 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 2.156e+02 2.695e+02 3.844e+02 9.084e+02, threshold=5.389e+02, percent-clipped=7.0 2022-12-08 09:42:41,603 INFO [train.py:873] (1/4) Epoch 16, batch 4600, loss[loss=0.104, simple_loss=0.1414, pruned_loss=0.03329, over 14465.00 frames. ], tot_loss[loss=0.1095, simple_loss=0.1454, pruned_loss=0.03682, over 2049366.90 frames. ], batch size: 49, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:42:43,361 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:42:50,010 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:43:42,510 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118101.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:43:45,464 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.071e+02 2.553e+02 3.122e+02 5.122e+02, threshold=5.107e+02, percent-clipped=0.0 2022-12-08 09:44:08,310 INFO [train.py:873] (1/4) Epoch 16, batch 4700, loss[loss=0.1572, simple_loss=0.1724, pruned_loss=0.07104, over 8599.00 frames. ], tot_loss[loss=0.1096, simple_loss=0.1453, pruned_loss=0.0369, over 2025750.78 frames. ], batch size: 100, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:44:19,653 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8844, 1.6662, 1.8648, 1.6454, 2.0110, 1.7586, 1.6914, 1.8758], device='cuda:1'), covar=tensor([0.0611, 0.1352, 0.0501, 0.0585, 0.0611, 0.0846, 0.0370, 0.0552], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0308, 0.0389, 0.0296, 0.0364, 0.0321, 0.0358, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:44:40,872 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118168.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:44:59,511 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 09:45:12,000 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 09:45:12,816 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2022-12-08 09:45:13,142 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.277e+02 2.708e+02 3.256e+02 5.977e+02, threshold=5.416e+02, percent-clipped=1.0 2022-12-08 09:45:27,792 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8999, 2.7250, 2.7801, 2.9236, 2.7831, 2.8466, 2.9960, 2.5578], device='cuda:1'), covar=tensor([0.0670, 0.1021, 0.0605, 0.0608, 0.0825, 0.0528, 0.0634, 0.0632], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0274, 0.0200, 0.0193, 0.0185, 0.0158, 0.0284, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:45:33,950 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118229.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:45:35,651 INFO [train.py:873] (1/4) Epoch 16, batch 4800, loss[loss=0.1236, simple_loss=0.1308, pruned_loss=0.05824, over 2636.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1445, pruned_loss=0.03677, over 1997567.60 frames. ], batch size: 100, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:46:08,460 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:46:26,144 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:46:39,568 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.128e+02 2.595e+02 3.127e+02 6.386e+02, threshold=5.189e+02, percent-clipped=3.0 2022-12-08 09:46:50,485 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 09:47:00,716 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:01,535 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118330.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:02,255 INFO [train.py:873] (1/4) Epoch 16, batch 4900, loss[loss=0.1299, simple_loss=0.1343, pruned_loss=0.06273, over 2561.00 frames. ], tot_loss[loss=0.1103, simple_loss=0.1454, pruned_loss=0.03763, over 1981263.35 frames. ], batch size: 100, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:47:18,682 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:39,405 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.2884, 4.8414, 4.7083, 5.2792, 4.8308, 4.4715, 5.2354, 4.4392], device='cuda:1'), covar=tensor([0.0313, 0.0827, 0.0403, 0.0378, 0.0818, 0.0553, 0.0491, 0.0492], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0273, 0.0198, 0.0192, 0.0184, 0.0157, 0.0283, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 09:47:53,190 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:58,365 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:48:06,579 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.054e+02 2.113e+02 2.635e+02 3.192e+02 6.518e+02, threshold=5.270e+02, percent-clipped=3.0 2022-12-08 09:48:29,343 INFO [train.py:873] (1/4) Epoch 16, batch 5000, loss[loss=0.1066, simple_loss=0.1382, pruned_loss=0.03746, over 14156.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1445, pruned_loss=0.03703, over 1985913.37 frames. ], batch size: 99, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:48:32,817 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0049, 4.7989, 4.5602, 4.9641, 4.5579, 4.3679, 5.0591, 4.7893], device='cuda:1'), covar=tensor([0.0632, 0.0896, 0.0805, 0.0587, 0.0742, 0.0491, 0.0545, 0.0711], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0142, 0.0146, 0.0159, 0.0146, 0.0123, 0.0168, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 09:49:02,997 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 09:49:34,275 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.516e+01 2.019e+02 2.454e+02 2.999e+02 6.382e+02, threshold=4.907e+02, percent-clipped=1.0 2022-12-08 09:49:38,543 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.67 vs. limit=5.0 2022-12-08 09:49:50,436 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118524.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:49:56,304 INFO [train.py:873] (1/4) Epoch 16, batch 5100, loss[loss=0.1082, simple_loss=0.143, pruned_loss=0.03673, over 14220.00 frames. ], tot_loss[loss=0.1094, simple_loss=0.1444, pruned_loss=0.03725, over 1973566.93 frames. ], batch size: 60, lr: 4.89e-03, grad_scale: 4.0 2022-12-08 09:50:26,947 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 09:51:01,826 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.028e+02 2.469e+02 3.033e+02 6.307e+02, threshold=4.938e+02, percent-clipped=1.0 2022-12-08 09:51:04,240 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2022-12-08 09:51:08,760 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9618, 4.8127, 4.3827, 4.5501, 4.7070, 4.8836, 4.9739, 4.9862], device='cuda:1'), covar=tensor([0.0765, 0.0368, 0.2049, 0.2584, 0.0641, 0.0774, 0.0713, 0.0667], device='cuda:1'), in_proj_covar=tensor([0.0390, 0.0269, 0.0449, 0.0561, 0.0348, 0.0447, 0.0391, 0.0389], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:51:18,453 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:51:23,926 INFO [train.py:873] (1/4) Epoch 16, batch 5200, loss[loss=0.09439, simple_loss=0.1385, pruned_loss=0.02513, over 14520.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.145, pruned_loss=0.0376, over 1990817.34 frames. ], batch size: 43, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:51:24,070 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118631.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:51:36,571 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118645.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:51:58,818 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 09:52:12,078 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:13,063 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0235, 2.1193, 4.0549, 2.8634, 3.8918, 2.0645, 3.0421, 3.9162], device='cuda:1'), covar=tensor([0.0644, 0.3744, 0.0445, 0.4968, 0.0665, 0.3077, 0.1304, 0.0487], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0204, 0.0217, 0.0276, 0.0237, 0.0206, 0.0205, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 09:52:18,137 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:21,220 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118696.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:29,670 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 1.977e+02 2.482e+02 3.057e+02 6.335e+02, threshold=4.964e+02, percent-clipped=4.0 2022-12-08 09:52:51,981 INFO [train.py:873] (1/4) Epoch 16, batch 5300, loss[loss=0.1082, simple_loss=0.1433, pruned_loss=0.03661, over 14293.00 frames. ], tot_loss[loss=0.1105, simple_loss=0.1456, pruned_loss=0.0377, over 1980540.84 frames. ], batch size: 25, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:53:03,350 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118744.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:53:22,418 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4978, 3.3870, 3.0617, 3.1979, 3.4748, 3.4373, 3.4861, 3.4821], device='cuda:1'), covar=tensor([0.0989, 0.0656, 0.2140, 0.2831, 0.0753, 0.1059, 0.1052, 0.0951], device='cuda:1'), in_proj_covar=tensor([0.0396, 0.0272, 0.0454, 0.0569, 0.0353, 0.0452, 0.0396, 0.0393], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 09:53:58,970 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 1.964e+02 2.394e+02 2.893e+02 4.670e+02, threshold=4.788e+02, percent-clipped=0.0 2022-12-08 09:54:14,761 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:54:20,951 INFO [train.py:873] (1/4) Epoch 16, batch 5400, loss[loss=0.1434, simple_loss=0.1665, pruned_loss=0.06019, over 8646.00 frames. ], tot_loss[loss=0.1108, simple_loss=0.1459, pruned_loss=0.03786, over 2014480.36 frames. ], batch size: 100, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:54:38,994 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2783, 2.8887, 3.8277, 2.6184, 2.4656, 2.9867, 1.8486, 3.4354], device='cuda:1'), covar=tensor([0.1059, 0.1191, 0.0543, 0.1541, 0.1895, 0.0716, 0.3041, 0.0734], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0102, 0.0095, 0.0101, 0.0116, 0.0090, 0.0121, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 09:54:57,514 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:55:07,338 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2010, 2.0343, 4.7454, 4.2738, 4.1127, 4.7824, 4.4518, 4.8491], device='cuda:1'), covar=tensor([0.1432, 0.1407, 0.0093, 0.0195, 0.0238, 0.0120, 0.0139, 0.0096], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0160, 0.0131, 0.0171, 0.0149, 0.0143, 0.0125, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 09:55:27,015 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.618e+01 2.141e+02 2.825e+02 3.414e+02 6.875e+02, threshold=5.649e+02, percent-clipped=6.0 2022-12-08 09:55:44,014 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:55:49,198 INFO [train.py:873] (1/4) Epoch 16, batch 5500, loss[loss=0.1233, simple_loss=0.161, pruned_loss=0.04281, over 14269.00 frames. ], tot_loss[loss=0.1092, simple_loss=0.1446, pruned_loss=0.03692, over 2000854.55 frames. ], batch size: 94, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:56:01,670 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:26,065 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:27,509 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 09:56:34,817 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1507, 1.2001, 1.0210, 1.1423, 1.1637, 0.7591, 1.0017, 1.1447], device='cuda:1'), covar=tensor([0.0633, 0.0649, 0.0495, 0.0407, 0.0443, 0.0733, 0.1011, 0.0609], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 09:56:36,590 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:38,461 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:43,684 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118993.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:55,520 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.181e+02 2.740e+02 3.306e+02 6.665e+02, threshold=5.480e+02, percent-clipped=3.0 2022-12-08 09:57:08,335 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.43 vs. limit=5.0 2022-12-08 09:57:17,516 INFO [train.py:873] (1/4) Epoch 16, batch 5600, loss[loss=0.0865, simple_loss=0.1372, pruned_loss=0.01788, over 14214.00 frames. ], tot_loss[loss=0.1099, simple_loss=0.145, pruned_loss=0.03741, over 2002738.11 frames. ], batch size: 32, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:57:19,251 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:57:21,253 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 09:57:55,304 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:58:11,118 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119093.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:58:21,974 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.186e+02 2.774e+02 3.458e+02 6.793e+02, threshold=5.547e+02, percent-clipped=4.0 2022-12-08 09:58:43,943 INFO [train.py:873] (1/4) Epoch 16, batch 5700, loss[loss=0.1514, simple_loss=0.1446, pruned_loss=0.07911, over 1244.00 frames. ], tot_loss[loss=0.1114, simple_loss=0.1458, pruned_loss=0.0385, over 1976820.76 frames. ], batch size: 100, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:58:48,195 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:59:04,230 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119154.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:59:50,092 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.314e+02 2.766e+02 3.252e+02 7.674e+02, threshold=5.531e+02, percent-clipped=3.0 2022-12-08 10:00:11,369 INFO [train.py:873] (1/4) Epoch 16, batch 5800, loss[loss=0.1031, simple_loss=0.1445, pruned_loss=0.03084, over 14266.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1452, pruned_loss=0.03814, over 1972404.29 frames. ], batch size: 39, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 10:00:14,336 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9827, 5.4863, 5.3954, 5.9995, 5.5867, 4.8138, 5.8881, 4.9150], device='cuda:1'), covar=tensor([0.0364, 0.0723, 0.0336, 0.0333, 0.0712, 0.0359, 0.0451, 0.0470], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0274, 0.0199, 0.0191, 0.0186, 0.0156, 0.0286, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:00:19,531 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7206, 4.5960, 4.3685, 4.7591, 4.4134, 4.0954, 4.8591, 4.5768], device='cuda:1'), covar=tensor([0.0570, 0.0740, 0.0722, 0.0482, 0.0643, 0.0648, 0.0441, 0.0695], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0143, 0.0146, 0.0159, 0.0146, 0.0123, 0.0167, 0.0148], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 10:00:24,206 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7557, 2.8563, 2.9658, 2.8871, 2.9019, 2.7633, 1.4899, 2.6815], device='cuda:1'), covar=tensor([0.0514, 0.0450, 0.0365, 0.0397, 0.0405, 0.0900, 0.2587, 0.0380], device='cuda:1'), in_proj_covar=tensor([0.0170, 0.0172, 0.0145, 0.0145, 0.0204, 0.0140, 0.0155, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 10:01:00,878 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:01:17,094 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.269e+02 2.799e+02 3.367e+02 7.852e+02, threshold=5.598e+02, percent-clipped=4.0 2022-12-08 10:01:38,987 INFO [train.py:873] (1/4) Epoch 16, batch 5900, loss[loss=0.1636, simple_loss=0.1462, pruned_loss=0.09053, over 1258.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.1449, pruned_loss=0.03768, over 1959360.27 frames. ], batch size: 100, lr: 4.88e-03, grad_scale: 4.0 2022-12-08 10:01:42,502 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:01:43,300 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7890, 2.6260, 2.2253, 2.4206, 2.7545, 2.7076, 2.7607, 2.7644], device='cuda:1'), covar=tensor([0.1266, 0.1030, 0.3130, 0.3375, 0.1325, 0.1539, 0.1522, 0.1184], device='cuda:1'), in_proj_covar=tensor([0.0393, 0.0269, 0.0451, 0.0566, 0.0350, 0.0448, 0.0393, 0.0388], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:02:00,587 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0949, 2.1404, 3.0231, 3.1540, 3.0429, 2.2782, 3.0746, 2.4612], device='cuda:1'), covar=tensor([0.0486, 0.1205, 0.0764, 0.0530, 0.0587, 0.1562, 0.0465, 0.1010], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0259, 0.0375, 0.0328, 0.0272, 0.0303, 0.0310, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 10:02:45,650 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.147e+02 2.622e+02 3.177e+02 5.866e+02, threshold=5.244e+02, percent-clipped=1.0 2022-12-08 10:02:47,312 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2022-12-08 10:03:05,590 INFO [train.py:873] (1/4) Epoch 16, batch 6000, loss[loss=0.1109, simple_loss=0.1332, pruned_loss=0.04425, over 4991.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.1444, pruned_loss=0.0367, over 1999974.36 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:03:05,591 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 10:03:14,045 INFO [train.py:905] (1/4) Epoch 16, validation: loss=0.1378, simple_loss=0.175, pruned_loss=0.05031, over 857387.00 frames. 2022-12-08 10:03:14,046 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 10:03:14,112 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119431.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:03:29,675 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119449.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:03:33,175 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4990, 3.8839, 3.0760, 4.8248, 4.2365, 4.5585, 3.9719, 3.4012], device='cuda:1'), covar=tensor([0.0896, 0.1231, 0.3310, 0.0501, 0.0963, 0.1283, 0.1121, 0.2707], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0292, 0.0262, 0.0285, 0.0322, 0.0300, 0.0255, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:04:20,754 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.903e+01 2.074e+02 2.533e+02 3.058e+02 7.007e+02, threshold=5.066e+02, percent-clipped=2.0 2022-12-08 10:04:26,269 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119514.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:04:41,253 INFO [train.py:873] (1/4) Epoch 16, batch 6100, loss[loss=0.09838, simple_loss=0.1118, pruned_loss=0.04249, over 2618.00 frames. ], tot_loss[loss=0.1096, simple_loss=0.1446, pruned_loss=0.03725, over 1942089.18 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:05:19,863 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:05:48,893 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 1.899e+02 2.518e+02 3.457e+02 7.704e+02, threshold=5.035e+02, percent-clipped=3.0 2022-12-08 10:06:09,183 INFO [train.py:873] (1/4) Epoch 16, batch 6200, loss[loss=0.1432, simple_loss=0.1647, pruned_loss=0.06082, over 7732.00 frames. ], tot_loss[loss=0.1096, simple_loss=0.145, pruned_loss=0.03708, over 2001013.37 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:06:33,205 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 10:06:38,600 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119665.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:06:50,889 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2022-12-08 10:07:16,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.165e+02 2.712e+02 3.301e+02 6.973e+02, threshold=5.424e+02, percent-clipped=2.0 2022-12-08 10:07:31,464 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8403, 1.8103, 2.1317, 1.8460, 1.9556, 1.6458, 1.5381, 1.2803], device='cuda:1'), covar=tensor([0.0254, 0.0532, 0.0199, 0.0265, 0.0229, 0.0312, 0.0299, 0.0547], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:07:32,924 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119726.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:07:37,349 INFO [train.py:873] (1/4) Epoch 16, batch 6300, loss[loss=0.1147, simple_loss=0.1569, pruned_loss=0.03625, over 14269.00 frames. ], tot_loss[loss=0.1083, simple_loss=0.144, pruned_loss=0.03635, over 2022055.34 frames. ], batch size: 76, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:07:37,447 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:07:53,285 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119749.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 10:08:19,739 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119779.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:08:35,911 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:08:46,035 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.194e+02 2.675e+02 3.528e+02 1.237e+03, threshold=5.350e+02, percent-clipped=2.0 2022-12-08 10:08:53,811 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6199, 4.7737, 5.0002, 4.3647, 4.7674, 5.1839, 1.9028, 4.4719], device='cuda:1'), covar=tensor([0.0297, 0.0293, 0.0354, 0.0386, 0.0303, 0.0132, 0.3064, 0.0292], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0174, 0.0146, 0.0144, 0.0203, 0.0140, 0.0157, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 10:09:05,868 INFO [train.py:873] (1/4) Epoch 16, batch 6400, loss[loss=0.118, simple_loss=0.1277, pruned_loss=0.05418, over 2655.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1441, pruned_loss=0.03657, over 1958527.77 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 8.0 2022-12-08 10:09:40,240 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:10:13,249 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.000e+02 2.417e+02 3.320e+02 6.714e+02, threshold=4.834e+02, percent-clipped=5.0 2022-12-08 10:10:33,939 INFO [train.py:873] (1/4) Epoch 16, batch 6500, loss[loss=0.08892, simple_loss=0.1356, pruned_loss=0.02111, over 14279.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1449, pruned_loss=0.03667, over 2006071.43 frames. ], batch size: 31, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:10:59,526 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1757, 2.1364, 4.7000, 2.7914, 4.6608, 5.2164, 4.9626, 5.5519], device='cuda:1'), covar=tensor([0.0173, 0.2560, 0.0447, 0.1672, 0.0270, 0.0376, 0.0254, 0.0170], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0159, 0.0167, 0.0166, 0.0179, 0.0132, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:11:44,555 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 1.947e+02 2.582e+02 3.066e+02 4.807e+02, threshold=5.164e+02, percent-clipped=0.0 2022-12-08 10:11:55,962 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=120021.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:12:04,535 INFO [train.py:873] (1/4) Epoch 16, batch 6600, loss[loss=0.1404, simple_loss=0.1402, pruned_loss=0.07027, over 1224.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1445, pruned_loss=0.03703, over 1936410.46 frames. ], batch size: 100, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:12:16,166 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8560, 1.8669, 1.6753, 1.9269, 1.7469, 1.8098, 1.8441, 1.7507], device='cuda:1'), covar=tensor([0.1298, 0.0780, 0.2044, 0.0820, 0.1205, 0.0706, 0.1210, 0.0925], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0291, 0.0260, 0.0284, 0.0321, 0.0299, 0.0252, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:12:24,214 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8368, 0.8614, 0.6760, 0.8887, 0.8389, 0.3654, 0.7858, 0.8763], device='cuda:1'), covar=tensor([0.0406, 0.0576, 0.0439, 0.0510, 0.0413, 0.0397, 0.1134, 0.0747], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0035, 0.0038, 0.0033, 0.0034, 0.0047, 0.0036, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:12:42,831 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 10:12:58,949 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9456, 3.7451, 3.6484, 4.0085, 3.7536, 3.4713, 4.0275, 3.4312], device='cuda:1'), covar=tensor([0.0631, 0.0919, 0.0465, 0.0440, 0.0764, 0.1675, 0.0572, 0.0545], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0273, 0.0198, 0.0191, 0.0183, 0.0155, 0.0284, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:13:12,085 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.190e+01 2.035e+02 2.583e+02 3.251e+02 5.143e+02, threshold=5.166e+02, percent-clipped=0.0 2022-12-08 10:13:32,124 INFO [train.py:873] (1/4) Epoch 16, batch 6700, loss[loss=0.1004, simple_loss=0.137, pruned_loss=0.03186, over 14234.00 frames. ], tot_loss[loss=0.1084, simple_loss=0.1438, pruned_loss=0.03653, over 1994832.45 frames. ], batch size: 69, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:13:34,157 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 10:13:52,949 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2022-12-08 10:14:06,012 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:14:08,959 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2022-12-08 10:14:39,089 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.165e+02 2.458e+02 2.899e+02 9.707e+02, threshold=4.916e+02, percent-clipped=4.0 2022-12-08 10:14:48,300 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=120218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:14:59,583 INFO [train.py:873] (1/4) Epoch 16, batch 6800, loss[loss=0.1374, simple_loss=0.1577, pruned_loss=0.05854, over 4942.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1431, pruned_loss=0.03577, over 2008826.26 frames. ], batch size: 100, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:16:06,647 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.118e+02 2.648e+02 3.259e+02 6.407e+02, threshold=5.296e+02, percent-clipped=3.0 2022-12-08 10:16:18,496 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:16:27,162 INFO [train.py:873] (1/4) Epoch 16, batch 6900, loss[loss=0.1138, simple_loss=0.1483, pruned_loss=0.03961, over 14283.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.144, pruned_loss=0.03612, over 2061134.54 frames. ], batch size: 69, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:16:28,995 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1152, 1.1996, 1.2997, 0.9841, 0.8519, 1.1680, 0.8432, 1.1786], device='cuda:1'), covar=tensor([0.2055, 0.3200, 0.1212, 0.2633, 0.3191, 0.1298, 0.1851, 0.1485], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0100, 0.0094, 0.0099, 0.0115, 0.0089, 0.0118, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 10:17:00,693 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=120369.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:17:34,669 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 2.083e+02 2.449e+02 3.128e+02 5.961e+02, threshold=4.898e+02, percent-clipped=1.0 2022-12-08 10:17:55,017 INFO [train.py:873] (1/4) Epoch 16, batch 7000, loss[loss=0.1049, simple_loss=0.1463, pruned_loss=0.03177, over 14355.00 frames. ], tot_loss[loss=0.1097, simple_loss=0.1447, pruned_loss=0.03733, over 1972546.22 frames. ], batch size: 73, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:18:47,249 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8692, 4.4806, 4.3214, 4.8529, 4.4783, 4.2156, 4.8571, 4.0991], device='cuda:1'), covar=tensor([0.0346, 0.0960, 0.0417, 0.0403, 0.0944, 0.0649, 0.0489, 0.0496], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0279, 0.0201, 0.0193, 0.0187, 0.0157, 0.0288, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:18:59,125 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7624, 5.2288, 5.1004, 5.7578, 5.2723, 4.7129, 5.6950, 4.7748], device='cuda:1'), covar=tensor([0.0269, 0.0963, 0.0406, 0.0332, 0.0919, 0.0370, 0.0427, 0.0466], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0279, 0.0201, 0.0194, 0.0187, 0.0157, 0.0288, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:19:01,657 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.060e+01 2.199e+02 2.820e+02 3.385e+02 6.936e+02, threshold=5.640e+02, percent-clipped=6.0 2022-12-08 10:19:21,574 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 10:19:21,716 INFO [train.py:873] (1/4) Epoch 16, batch 7100, loss[loss=0.1857, simple_loss=0.1579, pruned_loss=0.1067, over 1267.00 frames. ], tot_loss[loss=0.1097, simple_loss=0.1449, pruned_loss=0.03726, over 1999771.76 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:20:28,194 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.979e+01 2.056e+02 2.495e+02 3.157e+02 6.004e+02, threshold=4.990e+02, percent-clipped=1.0 2022-12-08 10:20:48,803 INFO [train.py:873] (1/4) Epoch 16, batch 7200, loss[loss=0.1463, simple_loss=0.1406, pruned_loss=0.076, over 2576.00 frames. ], tot_loss[loss=0.1102, simple_loss=0.1453, pruned_loss=0.03752, over 2025195.74 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:21:02,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2293, 2.2348, 3.0233, 2.4798, 3.0556, 3.0781, 2.9449, 2.6672], device='cuda:1'), covar=tensor([0.0863, 0.2871, 0.1046, 0.1844, 0.0867, 0.1051, 0.1045, 0.1730], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0314, 0.0395, 0.0303, 0.0371, 0.0325, 0.0362, 0.0303], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:21:27,306 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-08 10:21:27,868 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=120676.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:21:52,823 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9819, 1.9459, 1.9003, 1.9567, 1.9428, 1.1956, 1.7207, 1.8032], device='cuda:1'), covar=tensor([0.0719, 0.0542, 0.0570, 0.0824, 0.0763, 0.0787, 0.0656, 0.0613], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0035, 0.0038, 0.0033, 0.0034, 0.0047, 0.0036, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:21:55,119 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.056e+02 2.553e+02 2.947e+02 6.025e+02, threshold=5.105e+02, percent-clipped=4.0 2022-12-08 10:22:15,426 INFO [train.py:873] (1/4) Epoch 16, batch 7300, loss[loss=0.1104, simple_loss=0.148, pruned_loss=0.03647, over 14258.00 frames. ], tot_loss[loss=0.1104, simple_loss=0.1454, pruned_loss=0.03767, over 2065699.94 frames. ], batch size: 63, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:22:20,776 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=120737.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:23:22,447 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.181e+02 2.578e+02 3.347e+02 7.648e+02, threshold=5.157e+02, percent-clipped=5.0 2022-12-08 10:23:23,175 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 10:23:42,648 INFO [train.py:873] (1/4) Epoch 16, batch 7400, loss[loss=0.111, simple_loss=0.1467, pruned_loss=0.03767, over 14224.00 frames. ], tot_loss[loss=0.1109, simple_loss=0.1453, pruned_loss=0.03825, over 1999466.32 frames. ], batch size: 35, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:24:35,099 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5170, 4.2885, 4.0046, 4.1690, 4.3535, 4.4365, 4.5075, 4.4879], device='cuda:1'), covar=tensor([0.0774, 0.0494, 0.1878, 0.2521, 0.0688, 0.0816, 0.0820, 0.0798], device='cuda:1'), in_proj_covar=tensor([0.0393, 0.0274, 0.0452, 0.0572, 0.0350, 0.0447, 0.0393, 0.0393], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:24:49,561 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 2.203e+02 2.672e+02 3.215e+02 8.256e+02, threshold=5.345e+02, percent-clipped=4.0 2022-12-08 10:25:09,741 INFO [train.py:873] (1/4) Epoch 16, batch 7500, loss[loss=0.1094, simple_loss=0.1486, pruned_loss=0.03507, over 14265.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.145, pruned_loss=0.0381, over 1975836.02 frames. ], batch size: 80, lr: 4.84e-03, grad_scale: 8.0 2022-12-08 10:25:20,266 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3640, 2.7968, 4.1312, 3.1314, 4.2426, 4.0393, 4.0039, 3.5936], device='cuda:1'), covar=tensor([0.0726, 0.2610, 0.0944, 0.1742, 0.0765, 0.0892, 0.1340, 0.1532], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0312, 0.0392, 0.0301, 0.0369, 0.0322, 0.0359, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:25:32,268 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.23 vs. limit=5.0 2022-12-08 10:25:32,606 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0269, 4.1003, 4.3283, 3.7581, 4.1235, 4.3208, 1.7341, 3.9693], device='cuda:1'), covar=tensor([0.0314, 0.0353, 0.0308, 0.0525, 0.0322, 0.0267, 0.2925, 0.0268], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0175, 0.0146, 0.0147, 0.0206, 0.0142, 0.0159, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 10:25:49,737 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 10:26:35,449 INFO [train.py:873] (1/4) Epoch 17, batch 0, loss[loss=0.1206, simple_loss=0.1655, pruned_loss=0.03786, over 14287.00 frames. ], tot_loss[loss=0.1206, simple_loss=0.1655, pruned_loss=0.03786, over 14287.00 frames. ], batch size: 44, lr: 4.70e-03, grad_scale: 8.0 2022-12-08 10:26:35,450 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 10:26:39,059 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4972, 2.3494, 3.3672, 3.5251, 3.4541, 2.3216, 3.4421, 2.7523], device='cuda:1'), covar=tensor([0.0420, 0.1121, 0.0805, 0.0495, 0.0516, 0.1708, 0.0520, 0.1044], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0260, 0.0376, 0.0333, 0.0275, 0.0307, 0.0313, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 10:26:39,165 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.1565, 5.7050, 5.7458, 6.0687, 5.6796, 5.1232, 5.9659, 5.4691], device='cuda:1'), covar=tensor([0.0117, 0.0392, 0.0207, 0.0185, 0.0590, 0.0147, 0.0335, 0.0274], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0279, 0.0200, 0.0194, 0.0187, 0.0157, 0.0291, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:26:40,131 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8208, 1.9401, 1.8688, 2.0152, 1.8926, 1.2089, 1.6823, 1.7943], device='cuda:1'), covar=tensor([0.0727, 0.0518, 0.0482, 0.0809, 0.0479, 0.0846, 0.0782, 0.0802], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0037], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:26:42,839 INFO [train.py:905] (1/4) Epoch 17, validation: loss=0.1441, simple_loss=0.1813, pruned_loss=0.05348, over 857387.00 frames. 2022-12-08 10:26:42,840 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 10:26:52,031 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:26:56,390 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.236e+01 1.830e+02 2.606e+02 3.490e+02 1.014e+03, threshold=5.212e+02, percent-clipped=7.0 2022-12-08 10:27:02,241 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2517, 3.6748, 3.2637, 3.4455, 2.6055, 3.6166, 3.3748, 1.8557], device='cuda:1'), covar=tensor([0.1446, 0.0529, 0.1186, 0.0844, 0.0988, 0.0423, 0.0860, 0.2031], device='cuda:1'), in_proj_covar=tensor([0.0140, 0.0088, 0.0070, 0.0074, 0.0101, 0.0089, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:27:17,512 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:27:46,344 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121064.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:27:58,366 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 10:28:11,823 INFO [train.py:873] (1/4) Epoch 17, batch 100, loss[loss=0.09696, simple_loss=0.1419, pruned_loss=0.026, over 13941.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1458, pruned_loss=0.03612, over 887481.02 frames. ], batch size: 23, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:28:23,276 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0296, 1.2539, 1.3053, 1.0084, 0.7982, 1.1444, 0.8685, 1.2288], device='cuda:1'), covar=tensor([0.2486, 0.2959, 0.1438, 0.2603, 0.3812, 0.1335, 0.2172, 0.1429], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0094, 0.0100, 0.0117, 0.0090, 0.0119, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 10:28:24,821 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.352e+02 2.943e+02 3.420e+02 5.915e+02, threshold=5.886e+02, percent-clipped=2.0 2022-12-08 10:29:04,793 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6270, 1.6041, 1.7293, 1.5428, 1.4684, 1.6045, 1.4686, 1.1659], device='cuda:1'), covar=tensor([0.0169, 0.0292, 0.0173, 0.0197, 0.0237, 0.0310, 0.0249, 0.0395], device='cuda:1'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:29:13,765 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:29:39,133 INFO [train.py:873] (1/4) Epoch 17, batch 200, loss[loss=0.113, simple_loss=0.1527, pruned_loss=0.03664, over 14520.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1448, pruned_loss=0.03614, over 1356460.87 frames. ], batch size: 34, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:29:52,501 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.132e+02 2.532e+02 3.029e+02 6.779e+02, threshold=5.065e+02, percent-clipped=1.0 2022-12-08 10:30:07,057 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:30:17,729 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9908, 2.1195, 4.6522, 4.3134, 4.1336, 4.8178, 4.5912, 4.8376], device='cuda:1'), covar=tensor([0.1911, 0.1649, 0.0172, 0.0261, 0.0275, 0.0207, 0.0148, 0.0178], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0160, 0.0130, 0.0169, 0.0149, 0.0143, 0.0126, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 10:30:47,177 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 10:31:05,300 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 10:31:06,417 INFO [train.py:873] (1/4) Epoch 17, batch 300, loss[loss=0.11, simple_loss=0.1436, pruned_loss=0.03819, over 14244.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1449, pruned_loss=0.03678, over 1630627.30 frames. ], batch size: 80, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:31:19,308 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.023e+02 2.561e+02 3.068e+02 6.366e+02, threshold=5.121e+02, percent-clipped=2.0 2022-12-08 10:31:22,142 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121311.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:31:40,123 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121332.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:31:48,625 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-08 10:32:03,758 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121359.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:10,684 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8327, 2.4616, 3.6227, 2.8274, 3.6537, 3.5550, 3.4213, 3.0914], device='cuda:1'), covar=tensor([0.0803, 0.2883, 0.0953, 0.1799, 0.0915, 0.0986, 0.1345, 0.1507], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0311, 0.0388, 0.0299, 0.0367, 0.0320, 0.0359, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:32:14,970 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121372.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:32:22,002 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:33,192 INFO [train.py:873] (1/4) Epoch 17, batch 400, loss[loss=0.1126, simple_loss=0.1454, pruned_loss=0.03987, over 14634.00 frames. ], tot_loss[loss=0.1095, simple_loss=0.1449, pruned_loss=0.03703, over 1766926.27 frames. ], batch size: 33, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:32:45,742 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:47,376 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.084e+02 2.168e+02 2.580e+02 3.311e+02 5.353e+02, threshold=5.159e+02, percent-clipped=2.0 2022-12-08 10:33:28,398 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7258, 3.8136, 4.0049, 3.6412, 3.8957, 3.9453, 1.6544, 3.7060], device='cuda:1'), covar=tensor([0.0357, 0.0367, 0.0326, 0.0457, 0.0298, 0.0356, 0.2928, 0.0277], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0175, 0.0146, 0.0147, 0.0206, 0.0142, 0.0159, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 10:33:38,769 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:34:00,287 INFO [train.py:873] (1/4) Epoch 17, batch 500, loss[loss=0.1177, simple_loss=0.1483, pruned_loss=0.04357, over 14381.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1447, pruned_loss=0.0366, over 1899706.43 frames. ], batch size: 55, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:34:09,453 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0306, 2.0319, 2.0691, 2.1005, 2.0365, 1.6739, 1.2993, 1.8642], device='cuda:1'), covar=tensor([0.0661, 0.0576, 0.0465, 0.0390, 0.0469, 0.1331, 0.2067, 0.0506], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0175, 0.0146, 0.0146, 0.0206, 0.0142, 0.0159, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 10:34:14,595 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.218e+02 2.633e+02 3.024e+02 9.780e+02, threshold=5.266e+02, percent-clipped=2.0 2022-12-08 10:34:23,988 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:34:42,473 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3111, 2.8120, 4.1008, 3.1757, 4.1637, 4.1108, 3.9682, 3.4710], device='cuda:1'), covar=tensor([0.0739, 0.2689, 0.0910, 0.1561, 0.0760, 0.0836, 0.1274, 0.1671], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0310, 0.0389, 0.0298, 0.0364, 0.0320, 0.0358, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:34:54,316 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9065, 4.3016, 3.3595, 5.1243, 4.5469, 4.8740, 4.2540, 3.5927], device='cuda:1'), covar=tensor([0.0510, 0.0889, 0.2758, 0.0476, 0.0896, 0.1394, 0.0950, 0.2410], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0291, 0.0259, 0.0283, 0.0323, 0.0299, 0.0253, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:35:05,291 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1009, 2.1172, 2.4028, 1.5899, 1.6663, 2.1902, 1.2298, 2.2138], device='cuda:1'), covar=tensor([0.1009, 0.1402, 0.0722, 0.1912, 0.2327, 0.0920, 0.3226, 0.0911], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0102, 0.0096, 0.0100, 0.0118, 0.0091, 0.0120, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 10:35:11,941 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.24 vs. limit=5.0 2022-12-08 10:35:27,311 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 10:35:27,315 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2022-12-08 10:35:27,461 INFO [train.py:873] (1/4) Epoch 17, batch 600, loss[loss=0.1315, simple_loss=0.1344, pruned_loss=0.06432, over 2653.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1447, pruned_loss=0.03675, over 1912328.15 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:35:41,062 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.288e+02 2.719e+02 3.282e+02 7.294e+02, threshold=5.437e+02, percent-clipped=3.0 2022-12-08 10:36:25,588 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:36:32,570 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121667.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:36:32,688 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1810, 2.6848, 5.1816, 3.5981, 4.9350, 2.5707, 3.9755, 4.9770], device='cuda:1'), covar=tensor([0.0319, 0.3569, 0.0307, 0.5871, 0.0415, 0.3051, 0.1093, 0.0340], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0204, 0.0219, 0.0273, 0.0238, 0.0206, 0.0201, 0.0218], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 10:36:55,473 INFO [train.py:873] (1/4) Epoch 17, batch 700, loss[loss=0.09347, simple_loss=0.138, pruned_loss=0.02449, over 14561.00 frames. ], tot_loss[loss=0.1075, simple_loss=0.1434, pruned_loss=0.0358, over 1926008.35 frames. ], batch size: 34, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:37:08,201 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:37:09,978 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.051e+02 2.543e+02 3.291e+02 5.675e+02, threshold=5.087e+02, percent-clipped=1.0 2022-12-08 10:37:19,882 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.49 vs. limit=2.0 2022-12-08 10:37:52,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3444, 0.9986, 1.2224, 0.7964, 1.2005, 1.3506, 1.0320, 1.1221], device='cuda:1'), covar=tensor([0.0595, 0.0900, 0.0895, 0.0590, 0.1302, 0.0799, 0.0762, 0.1401], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0170, 0.0141, 0.0126, 0.0143, 0.0154, 0.0133, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:37:57,031 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:38:23,111 INFO [train.py:873] (1/4) Epoch 17, batch 800, loss[loss=0.1218, simple_loss=0.1227, pruned_loss=0.06048, over 2704.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1438, pruned_loss=0.03659, over 1926035.30 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:38:37,353 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.062e+02 2.469e+02 3.248e+02 9.190e+02, threshold=4.938e+02, percent-clipped=3.0 2022-12-08 10:38:38,192 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-08 10:38:47,577 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:38:55,601 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9633, 3.5345, 2.7265, 4.1576, 3.9368, 4.0367, 3.4600, 2.8258], device='cuda:1'), covar=tensor([0.0753, 0.1159, 0.3048, 0.0606, 0.0811, 0.0927, 0.1192, 0.3018], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0293, 0.0260, 0.0285, 0.0326, 0.0301, 0.0255, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:39:02,027 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2022-12-08 10:39:02,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8872, 2.7499, 2.4903, 2.6369, 2.8047, 2.8395, 2.8469, 2.8746], device='cuda:1'), covar=tensor([0.1130, 0.0729, 0.2448, 0.2799, 0.1029, 0.1190, 0.1438, 0.0942], device='cuda:1'), in_proj_covar=tensor([0.0392, 0.0268, 0.0447, 0.0563, 0.0347, 0.0445, 0.0393, 0.0389], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:39:12,768 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1976, 4.0786, 3.9108, 4.2601, 3.8195, 3.5835, 4.3001, 4.0074], device='cuda:1'), covar=tensor([0.0623, 0.0863, 0.0783, 0.0565, 0.1005, 0.0697, 0.0553, 0.0814], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0143, 0.0146, 0.0159, 0.0147, 0.0122, 0.0167, 0.0147], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 10:39:29,520 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121868.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:39:51,404 INFO [train.py:873] (1/4) Epoch 17, batch 900, loss[loss=0.105, simple_loss=0.142, pruned_loss=0.03396, over 6949.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.1434, pruned_loss=0.03607, over 1985895.83 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:39:52,337 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8828, 0.8747, 0.9201, 0.8616, 0.7532, 0.6048, 0.7488, 0.7801], device='cuda:1'), covar=tensor([0.0295, 0.0277, 0.0248, 0.0276, 0.0303, 0.0458, 0.0349, 0.0491], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0021, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:40:04,967 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.025e+02 2.484e+02 3.273e+02 4.983e+02, threshold=4.968e+02, percent-clipped=1.0 2022-12-08 10:40:22,431 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2528, 3.9008, 3.5581, 3.5295, 2.7686, 3.7395, 3.4902, 2.0429], device='cuda:1'), covar=tensor([0.1554, 0.0423, 0.1110, 0.0654, 0.0893, 0.0405, 0.1005, 0.1838], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0089, 0.0069, 0.0075, 0.0100, 0.0088, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:40:40,928 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:40:40,975 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9233, 2.3875, 3.8076, 2.7062, 3.8171, 3.6816, 3.5923, 3.2727], device='cuda:1'), covar=tensor([0.0835, 0.3308, 0.0977, 0.2000, 0.0904, 0.0953, 0.1422, 0.1820], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0313, 0.0393, 0.0301, 0.0367, 0.0325, 0.0361, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:40:53,709 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0086, 2.6292, 3.4950, 2.4639, 2.2034, 2.8700, 1.9110, 3.1502], device='cuda:1'), covar=tensor([0.0961, 0.1341, 0.0651, 0.1986, 0.2103, 0.1111, 0.2906, 0.0712], device='cuda:1'), in_proj_covar=tensor([0.0084, 0.0100, 0.0095, 0.0098, 0.0115, 0.0090, 0.0118, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 10:40:54,552 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9605, 1.7713, 3.9720, 3.6715, 3.7724, 4.0728, 3.5109, 4.0207], device='cuda:1'), covar=tensor([0.1523, 0.1553, 0.0135, 0.0255, 0.0245, 0.0151, 0.0225, 0.0151], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0159, 0.0131, 0.0170, 0.0149, 0.0145, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 10:40:55,427 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121967.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:41:17,348 INFO [train.py:873] (1/4) Epoch 17, batch 1000, loss[loss=0.1157, simple_loss=0.1483, pruned_loss=0.04159, over 10381.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1433, pruned_loss=0.03655, over 1914375.04 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:41:31,554 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 2.007e+02 2.510e+02 3.013e+02 6.170e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 10:41:33,513 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122011.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:41:36,914 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122015.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:41:56,296 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9403, 2.1857, 2.2539, 2.3651, 2.0009, 2.3021, 2.0580, 1.4951], device='cuda:1'), covar=tensor([0.0820, 0.1019, 0.0678, 0.0568, 0.1064, 0.0669, 0.1227, 0.1916], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0089, 0.0069, 0.0075, 0.0100, 0.0088, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:42:01,767 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3431, 2.3766, 2.5144, 2.0721, 2.5259, 2.2200, 2.2164, 2.1878], device='cuda:1'), covar=tensor([0.0246, 0.0562, 0.0309, 0.0444, 0.0338, 0.0344, 0.0373, 0.0443], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 10:42:06,476 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 10:42:18,897 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:42:20,019 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 10:42:44,897 INFO [train.py:873] (1/4) Epoch 17, batch 1100, loss[loss=0.1272, simple_loss=0.159, pruned_loss=0.04767, over 14382.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1431, pruned_loss=0.03636, over 1958020.64 frames. ], batch size: 73, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:42:46,302 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-08 10:42:59,064 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.098e+02 2.553e+02 3.146e+02 5.520e+02, threshold=5.106e+02, percent-clipped=3.0 2022-12-08 10:43:00,810 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122111.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:43:05,850 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:43:59,278 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122178.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 10:44:12,205 INFO [train.py:873] (1/4) Epoch 17, batch 1200, loss[loss=0.1022, simple_loss=0.1427, pruned_loss=0.03088, over 14468.00 frames. ], tot_loss[loss=0.1094, simple_loss=0.1446, pruned_loss=0.03707, over 1982471.59 frames. ], batch size: 51, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:44:13,780 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122195.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:44:15,491 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3671, 1.0891, 1.2708, 0.8542, 1.1754, 1.4188, 1.0953, 1.1383], device='cuda:1'), covar=tensor([0.0461, 0.0781, 0.0838, 0.0512, 0.1198, 0.0806, 0.0597, 0.1438], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0170, 0.0139, 0.0126, 0.0143, 0.0153, 0.0133, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:44:19,713 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 10:44:26,232 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.994e+02 2.399e+02 3.114e+02 6.684e+02, threshold=4.798e+02, percent-clipped=5.0 2022-12-08 10:44:45,189 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2932, 2.2959, 1.8047, 2.2920, 2.2242, 2.2156, 2.1191, 2.0149], device='cuda:1'), covar=tensor([0.0989, 0.0965, 0.1789, 0.1114, 0.1160, 0.0840, 0.1511, 0.1463], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0289, 0.0257, 0.0281, 0.0320, 0.0298, 0.0252, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:45:07,632 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122256.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:45:39,706 INFO [train.py:873] (1/4) Epoch 17, batch 1300, loss[loss=0.1281, simple_loss=0.155, pruned_loss=0.0506, over 9513.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1437, pruned_loss=0.03681, over 1945633.77 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:45:52,011 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122306.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:45:54,349 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.064e+02 2.522e+02 2.867e+02 6.773e+02, threshold=5.044e+02, percent-clipped=2.0 2022-12-08 10:46:03,022 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:46:15,076 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.43 vs. limit=2.0 2022-12-08 10:46:15,641 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0243, 1.5807, 3.9457, 1.6991, 3.9438, 4.1246, 3.3500, 4.4544], device='cuda:1'), covar=tensor([0.0249, 0.3308, 0.0484, 0.2523, 0.0414, 0.0451, 0.0600, 0.0169], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0156, 0.0160, 0.0169, 0.0168, 0.0180, 0.0132, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:46:22,535 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 10:46:37,574 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8956, 3.7738, 3.4284, 3.5658, 3.8274, 3.8483, 3.8921, 3.9165], device='cuda:1'), covar=tensor([0.0875, 0.0487, 0.2005, 0.2463, 0.0709, 0.0777, 0.0909, 0.0700], device='cuda:1'), in_proj_covar=tensor([0.0385, 0.0266, 0.0444, 0.0560, 0.0344, 0.0439, 0.0389, 0.0387], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 10:46:49,433 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 10:46:52,886 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 10:46:56,002 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:47:07,174 INFO [train.py:873] (1/4) Epoch 17, batch 1400, loss[loss=0.111, simple_loss=0.1508, pruned_loss=0.0356, over 14260.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1441, pruned_loss=0.03695, over 1906115.09 frames. ], batch size: 28, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:47:21,304 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.812e+01 2.172e+02 2.630e+02 3.463e+02 6.435e+02, threshold=5.259e+02, percent-clipped=4.0 2022-12-08 10:47:29,493 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.46 vs. limit=5.0 2022-12-08 10:47:44,411 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 10:48:13,704 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:48:17,828 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122473.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 10:48:35,289 INFO [train.py:873] (1/4) Epoch 17, batch 1500, loss[loss=0.1242, simple_loss=0.1379, pruned_loss=0.05523, over 2582.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.1435, pruned_loss=0.03711, over 1875062.39 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:48:49,379 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.116e+02 2.482e+02 3.540e+02 1.353e+03, threshold=4.965e+02, percent-clipped=3.0 2022-12-08 10:49:06,546 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:49:26,301 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:50:03,166 INFO [train.py:873] (1/4) Epoch 17, batch 1600, loss[loss=0.1103, simple_loss=0.1251, pruned_loss=0.04773, over 2610.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1436, pruned_loss=0.03677, over 1937702.58 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:50:14,880 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122606.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:50:17,232 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.386e+01 1.886e+02 2.427e+02 3.060e+02 1.117e+03, threshold=4.854e+02, percent-clipped=4.0 2022-12-08 10:50:56,450 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122654.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:51:15,297 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122675.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:51:30,440 INFO [train.py:873] (1/4) Epoch 17, batch 1700, loss[loss=0.1279, simple_loss=0.1533, pruned_loss=0.05128, over 9487.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1437, pruned_loss=0.03683, over 1971748.25 frames. ], batch size: 100, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:51:45,734 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 2.028e+02 2.462e+02 2.959e+02 4.889e+02, threshold=4.923e+02, percent-clipped=1.0 2022-12-08 10:52:40,037 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122773.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:52:45,354 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8689, 1.4772, 2.9956, 2.6914, 2.8713, 3.0336, 2.3181, 2.9719], device='cuda:1'), covar=tensor([0.1296, 0.1532, 0.0187, 0.0422, 0.0376, 0.0215, 0.0556, 0.0223], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0159, 0.0131, 0.0168, 0.0148, 0.0143, 0.0124, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 10:52:45,883 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 10:52:57,841 INFO [train.py:873] (1/4) Epoch 17, batch 1800, loss[loss=0.1523, simple_loss=0.1492, pruned_loss=0.07765, over 1171.00 frames. ], tot_loss[loss=0.1092, simple_loss=0.1441, pruned_loss=0.03713, over 1951541.09 frames. ], batch size: 100, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:53:12,808 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 2.194e+02 2.593e+02 3.400e+02 7.475e+02, threshold=5.185e+02, percent-clipped=4.0 2022-12-08 10:53:22,454 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:53:24,998 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:53:48,793 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122851.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:54:07,225 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:54:25,551 INFO [train.py:873] (1/4) Epoch 17, batch 1900, loss[loss=0.1027, simple_loss=0.1427, pruned_loss=0.03135, over 14566.00 frames. ], tot_loss[loss=0.1095, simple_loss=0.1443, pruned_loss=0.03738, over 1962571.76 frames. ], batch size: 23, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:54:30,920 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:54:40,367 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.015e+02 2.576e+02 3.102e+02 5.463e+02, threshold=5.151e+02, percent-clipped=2.0 2022-12-08 10:55:00,444 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:55:37,062 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122975.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:55:40,483 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7259, 3.4584, 3.3710, 3.6888, 3.4928, 3.6671, 3.7304, 3.0852], device='cuda:1'), covar=tensor([0.0418, 0.0958, 0.0480, 0.0468, 0.0743, 0.0330, 0.0604, 0.0599], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0278, 0.0201, 0.0195, 0.0185, 0.0157, 0.0289, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 10:55:52,710 INFO [train.py:873] (1/4) Epoch 17, batch 2000, loss[loss=0.1022, simple_loss=0.1403, pruned_loss=0.03205, over 14238.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1445, pruned_loss=0.03654, over 2031641.06 frames. ], batch size: 89, lr: 4.66e-03, grad_scale: 8.0 2022-12-08 10:56:08,073 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 2.103e+02 2.584e+02 3.281e+02 5.069e+02, threshold=5.168e+02, percent-clipped=0.0 2022-12-08 10:56:09,925 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4661, 3.8246, 3.2421, 3.4988, 2.8672, 3.6967, 3.5521, 2.1184], device='cuda:1'), covar=tensor([0.1107, 0.0511, 0.1692, 0.0895, 0.0794, 0.0352, 0.0862, 0.1704], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0088, 0.0069, 0.0074, 0.0099, 0.0088, 0.0100, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 10:56:19,430 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123023.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:56:24,030 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 10:57:20,549 INFO [train.py:873] (1/4) Epoch 17, batch 2100, loss[loss=0.1316, simple_loss=0.1478, pruned_loss=0.05772, over 5959.00 frames. ], tot_loss[loss=0.1084, simple_loss=0.1437, pruned_loss=0.03653, over 1997359.97 frames. ], batch size: 100, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:57:24,606 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 10:57:36,150 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.207e+02 2.592e+02 3.533e+02 7.488e+02, threshold=5.184e+02, percent-clipped=8.0 2022-12-08 10:57:47,115 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:58:11,472 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-08 10:58:29,177 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:58:46,984 INFO [train.py:873] (1/4) Epoch 17, batch 2200, loss[loss=0.2004, simple_loss=0.1755, pruned_loss=0.1127, over 1256.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1443, pruned_loss=0.037, over 1991694.54 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 4.0 2022-12-08 10:59:03,099 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.073e+02 2.495e+02 2.880e+02 5.648e+02, threshold=4.991e+02, percent-clipped=1.0 2022-12-08 10:59:06,148 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.41 vs. limit=5.0 2022-12-08 10:59:17,842 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:00:13,630 INFO [train.py:873] (1/4) Epoch 17, batch 2300, loss[loss=0.1248, simple_loss=0.1247, pruned_loss=0.06246, over 1284.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1441, pruned_loss=0.03675, over 1955508.36 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 4.0 2022-12-08 11:00:29,910 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.209e+02 2.627e+02 3.184e+02 5.736e+02, threshold=5.255e+02, percent-clipped=2.0 2022-12-08 11:00:32,685 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123314.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:01:09,123 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-08 11:01:26,401 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123375.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:01:41,811 INFO [train.py:873] (1/4) Epoch 17, batch 2400, loss[loss=0.09218, simple_loss=0.133, pruned_loss=0.02568, over 14642.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.144, pruned_loss=0.03662, over 1946455.13 frames. ], batch size: 33, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:01:54,497 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2022-12-08 11:01:58,032 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.088e+02 2.563e+02 3.123e+02 4.912e+02, threshold=5.125e+02, percent-clipped=0.0 2022-12-08 11:02:12,584 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:02:13,463 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1714, 3.9362, 3.6565, 3.8141, 4.0449, 4.0835, 4.1573, 4.1594], device='cuda:1'), covar=tensor([0.0879, 0.0617, 0.2050, 0.2698, 0.0707, 0.0884, 0.0906, 0.0800], device='cuda:1'), in_proj_covar=tensor([0.0387, 0.0269, 0.0448, 0.0565, 0.0344, 0.0445, 0.0392, 0.0393], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:02:38,582 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3342, 3.7526, 2.9619, 4.6508, 4.1163, 4.3504, 3.9208, 3.1891], device='cuda:1'), covar=tensor([0.0813, 0.1228, 0.3417, 0.0488, 0.1027, 0.1653, 0.1140, 0.2766], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0290, 0.0258, 0.0285, 0.0321, 0.0300, 0.0252, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:03:07,264 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:03:10,657 INFO [train.py:873] (1/4) Epoch 17, batch 2500, loss[loss=0.109, simple_loss=0.1455, pruned_loss=0.03628, over 14420.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1436, pruned_loss=0.03615, over 2014738.65 frames. ], batch size: 73, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:03:26,840 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 2.132e+02 2.678e+02 3.392e+02 6.335e+02, threshold=5.356e+02, percent-clipped=3.0 2022-12-08 11:03:42,034 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:03:44,821 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1226, 2.8385, 2.9454, 2.1300, 2.6575, 2.8856, 3.1424, 2.6022], device='cuda:1'), covar=tensor([0.0722, 0.0825, 0.0736, 0.1244, 0.0841, 0.0796, 0.0634, 0.1087], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0171, 0.0140, 0.0126, 0.0143, 0.0155, 0.0135, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:03:59,464 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5452, 1.4495, 1.4086, 1.6037, 1.6564, 1.0284, 1.4037, 1.3806], device='cuda:1'), covar=tensor([0.0619, 0.0707, 0.0654, 0.0501, 0.0520, 0.0929, 0.0737, 0.0685], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0034, 0.0038, 0.0031, 0.0033, 0.0046, 0.0035, 0.0037], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 11:04:23,966 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:04:37,298 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2035, 3.9056, 3.6868, 3.8278, 4.0565, 4.0868, 4.1546, 4.1410], device='cuda:1'), covar=tensor([0.0751, 0.0592, 0.2051, 0.2515, 0.0706, 0.0817, 0.0977, 0.0806], device='cuda:1'), in_proj_covar=tensor([0.0389, 0.0272, 0.0452, 0.0566, 0.0346, 0.0446, 0.0392, 0.0396], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:04:39,204 INFO [train.py:873] (1/4) Epoch 17, batch 2600, loss[loss=0.1049, simple_loss=0.128, pruned_loss=0.0409, over 2624.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1438, pruned_loss=0.03627, over 2022934.90 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:04:43,615 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5035, 3.2659, 3.0525, 3.2061, 3.4207, 3.4421, 3.4511, 3.4909], device='cuda:1'), covar=tensor([0.0977, 0.0734, 0.2155, 0.2631, 0.0810, 0.0860, 0.1281, 0.0832], device='cuda:1'), in_proj_covar=tensor([0.0389, 0.0272, 0.0452, 0.0567, 0.0346, 0.0446, 0.0392, 0.0396], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:04:52,794 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6494, 4.4171, 4.1291, 4.2506, 4.4922, 4.5554, 4.6410, 4.6537], device='cuda:1'), covar=tensor([0.0871, 0.0559, 0.2006, 0.2803, 0.0671, 0.0811, 0.0864, 0.0765], device='cuda:1'), in_proj_covar=tensor([0.0390, 0.0273, 0.0453, 0.0568, 0.0347, 0.0447, 0.0393, 0.0397], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:04:55,081 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 1.985e+02 2.407e+02 3.072e+02 5.323e+02, threshold=4.814e+02, percent-clipped=0.0 2022-12-08 11:05:06,424 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0815, 2.5232, 2.7336, 1.8051, 2.5745, 2.8285, 2.9892, 2.3825], device='cuda:1'), covar=tensor([0.0764, 0.1029, 0.1002, 0.1669, 0.0951, 0.0716, 0.0914, 0.1489], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0171, 0.0141, 0.0127, 0.0143, 0.0155, 0.0135, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:05:12,515 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9197, 1.2533, 4.1959, 1.8456, 3.9573, 4.3828, 3.7859, 4.2888], device='cuda:1'), covar=tensor([0.0499, 0.4988, 0.0740, 0.3546, 0.0638, 0.0624, 0.0569, 0.0650], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0159, 0.0168, 0.0166, 0.0178, 0.0132, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:05:46,079 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123670.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:06:06,727 INFO [train.py:873] (1/4) Epoch 17, batch 2700, loss[loss=0.09854, simple_loss=0.1408, pruned_loss=0.02815, over 14270.00 frames. ], tot_loss[loss=0.1083, simple_loss=0.1439, pruned_loss=0.03633, over 2016117.02 frames. ], batch size: 76, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:06:22,182 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.020e+02 2.571e+02 3.169e+02 5.832e+02, threshold=5.143e+02, percent-clipped=4.0 2022-12-08 11:07:25,920 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:07:33,392 INFO [train.py:873] (1/4) Epoch 17, batch 2800, loss[loss=0.1754, simple_loss=0.157, pruned_loss=0.09684, over 1232.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1436, pruned_loss=0.03622, over 1968218.33 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:07:34,855 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7710, 2.9129, 4.5277, 3.5724, 4.5231, 4.3335, 4.3855, 3.9879], device='cuda:1'), covar=tensor([0.0666, 0.3022, 0.1076, 0.1590, 0.0725, 0.0956, 0.1354, 0.1633], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0311, 0.0393, 0.0299, 0.0366, 0.0322, 0.0360, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:07:49,825 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.052e+02 2.657e+02 3.438e+02 8.382e+02, threshold=5.313e+02, percent-clipped=4.0 2022-12-08 11:07:56,620 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9035, 4.0418, 4.0425, 3.8578, 3.9685, 4.1024, 1.6467, 3.7555], device='cuda:1'), covar=tensor([0.0482, 0.0517, 0.0557, 0.0613, 0.0554, 0.0387, 0.3798, 0.0491], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0176, 0.0147, 0.0148, 0.0208, 0.0143, 0.0158, 0.0195], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 11:08:42,954 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:08:50,769 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6720, 2.3258, 3.4313, 2.6854, 3.3957, 3.3560, 3.2376, 2.8765], device='cuda:1'), covar=tensor([0.0850, 0.3154, 0.1090, 0.1937, 0.0810, 0.1109, 0.1319, 0.1681], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0313, 0.0394, 0.0300, 0.0368, 0.0322, 0.0361, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:09:01,387 INFO [train.py:873] (1/4) Epoch 17, batch 2900, loss[loss=0.1143, simple_loss=0.1488, pruned_loss=0.03988, over 12791.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1436, pruned_loss=0.03634, over 1961868.66 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:09:16,751 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.204e+02 2.576e+02 3.173e+02 6.433e+02, threshold=5.152e+02, percent-clipped=4.0 2022-12-08 11:09:36,625 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:09:52,112 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5364, 1.4025, 2.7457, 1.5294, 2.7472, 2.7220, 2.0822, 2.9061], device='cuda:1'), covar=tensor([0.0282, 0.2819, 0.0408, 0.1996, 0.0358, 0.0472, 0.1125, 0.0260], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0154, 0.0158, 0.0167, 0.0165, 0.0177, 0.0132, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:09:55,432 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 11:09:59,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 11:10:08,162 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123970.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:10:27,939 INFO [train.py:873] (1/4) Epoch 17, batch 3000, loss[loss=0.08578, simple_loss=0.1236, pruned_loss=0.02398, over 5944.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1436, pruned_loss=0.03638, over 1945627.84 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:10:27,939 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 11:10:36,502 INFO [train.py:905] (1/4) Epoch 17, validation: loss=0.1392, simple_loss=0.1759, pruned_loss=0.05127, over 857387.00 frames. 2022-12-08 11:10:36,502 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 11:10:42,675 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3756, 3.2353, 3.9762, 2.8230, 2.3522, 3.4198, 1.8138, 3.5126], device='cuda:1'), covar=tensor([0.1063, 0.0934, 0.0548, 0.1842, 0.2134, 0.0720, 0.3146, 0.1160], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0099, 0.0094, 0.0099, 0.0114, 0.0089, 0.0117, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 11:10:52,778 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.049e+02 2.779e+02 3.287e+02 7.415e+02, threshold=5.559e+02, percent-clipped=4.0 2022-12-08 11:10:58,410 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:14,555 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:37,610 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124062.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:57,165 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:04,346 INFO [train.py:873] (1/4) Epoch 17, batch 3100, loss[loss=0.1222, simple_loss=0.152, pruned_loss=0.04619, over 14389.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1438, pruned_loss=0.03668, over 1961892.12 frames. ], batch size: 73, lr: 4.64e-03, grad_scale: 4.0 2022-12-08 11:12:08,231 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:20,720 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.198e+02 2.691e+02 3.318e+02 1.640e+03, threshold=5.381e+02, percent-clipped=4.0 2022-12-08 11:12:30,993 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124123.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:38,790 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:48,614 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.0849, 1.2851, 1.3045, 0.9975, 0.8196, 1.1368, 0.8853, 1.2688], device='cuda:1'), covar=tensor([0.2208, 0.2893, 0.1338, 0.3098, 0.3654, 0.1381, 0.2085, 0.1481], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0099, 0.0094, 0.0098, 0.0114, 0.0089, 0.0116, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 11:13:21,773 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0104, 3.7442, 3.5381, 3.7024, 3.8988, 3.9163, 3.9849, 4.0079], device='cuda:1'), covar=tensor([0.0865, 0.0633, 0.2043, 0.2265, 0.0730, 0.0793, 0.1003, 0.0785], device='cuda:1'), in_proj_covar=tensor([0.0390, 0.0273, 0.0453, 0.0567, 0.0347, 0.0448, 0.0393, 0.0399], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:13:31,917 INFO [train.py:873] (1/4) Epoch 17, batch 3200, loss[loss=0.093, simple_loss=0.1364, pruned_loss=0.02479, over 14597.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1439, pruned_loss=0.03652, over 1939627.54 frames. ], batch size: 30, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:13:43,890 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0038, 3.4487, 2.6447, 4.2019, 3.9480, 4.0071, 3.4105, 2.8646], device='cuda:1'), covar=tensor([0.0768, 0.1233, 0.3291, 0.0672, 0.0931, 0.1524, 0.1225, 0.3127], device='cuda:1'), in_proj_covar=tensor([0.0285, 0.0292, 0.0262, 0.0290, 0.0326, 0.0304, 0.0258, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:13:48,754 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 1.975e+02 2.521e+02 3.067e+02 5.500e+02, threshold=5.042e+02, percent-clipped=1.0 2022-12-08 11:14:02,434 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:14:40,346 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:14:50,119 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4979, 2.2780, 4.4805, 3.0019, 4.3533, 1.9903, 3.3539, 4.3026], device='cuda:1'), covar=tensor([0.0486, 0.3907, 0.0376, 0.5571, 0.0501, 0.3582, 0.1315, 0.0363], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0200, 0.0217, 0.0270, 0.0235, 0.0203, 0.0201, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 11:14:58,460 INFO [train.py:873] (1/4) Epoch 17, batch 3300, loss[loss=0.1183, simple_loss=0.1524, pruned_loss=0.04204, over 14127.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1438, pruned_loss=0.03679, over 1924032.81 frames. ], batch size: 99, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:15:14,505 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.029e+02 2.735e+02 3.344e+02 8.067e+02, threshold=5.469e+02, percent-clipped=3.0 2022-12-08 11:15:33,619 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124333.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:15:47,143 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124348.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 11:16:02,930 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5109, 1.8284, 1.9030, 2.0548, 1.8116, 2.0677, 1.7745, 1.1962], device='cuda:1'), covar=tensor([0.0821, 0.1033, 0.1043, 0.0720, 0.1127, 0.0749, 0.1536, 0.2430], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0089, 0.0071, 0.0076, 0.0100, 0.0089, 0.0102, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:16:25,320 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124392.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:16:26,083 INFO [train.py:873] (1/4) Epoch 17, batch 3400, loss[loss=0.1001, simple_loss=0.1365, pruned_loss=0.03184, over 14557.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1437, pruned_loss=0.03695, over 1928319.75 frames. ], batch size: 34, lr: 4.63e-03, grad_scale: 4.0 2022-12-08 11:16:40,735 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124409.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:16:43,978 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.119e+02 2.705e+02 3.387e+02 6.415e+02, threshold=5.409e+02, percent-clipped=2.0 2022-12-08 11:16:48,644 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124418.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:17:00,975 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:17:06,636 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8157, 1.6261, 1.8051, 1.9860, 1.5121, 1.6628, 1.7005, 1.8433], device='cuda:1'), covar=tensor([0.0212, 0.0356, 0.0196, 0.0170, 0.0324, 0.0473, 0.0283, 0.0197], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0256, 0.0373, 0.0329, 0.0270, 0.0305, 0.0312, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:17:47,119 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1348, 2.1989, 3.0153, 3.2109, 3.0405, 2.1805, 3.1066, 2.4503], device='cuda:1'), covar=tensor([0.0523, 0.1270, 0.0866, 0.0547, 0.0649, 0.1715, 0.0527, 0.1132], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0256, 0.0375, 0.0329, 0.0271, 0.0306, 0.0313, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:17:54,879 INFO [train.py:873] (1/4) Epoch 17, batch 3500, loss[loss=0.1011, simple_loss=0.1418, pruned_loss=0.03018, over 14221.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.144, pruned_loss=0.03648, over 1931396.82 frames. ], batch size: 60, lr: 4.63e-03, grad_scale: 4.0 2022-12-08 11:17:55,103 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:18:08,088 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8891, 1.7056, 1.8371, 1.6475, 1.9921, 1.7685, 1.5658, 1.8714], device='cuda:1'), covar=tensor([0.0711, 0.1279, 0.0475, 0.0653, 0.0474, 0.0855, 0.0330, 0.0460], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0313, 0.0395, 0.0302, 0.0368, 0.0324, 0.0364, 0.0300], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:18:12,216 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.179e+02 2.842e+02 3.919e+02 2.207e+03, threshold=5.685e+02, percent-clipped=7.0 2022-12-08 11:18:14,822 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 11:18:25,930 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:18:36,096 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-08 11:19:08,274 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:19:18,878 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2022-12-08 11:19:22,686 INFO [train.py:873] (1/4) Epoch 17, batch 3600, loss[loss=0.147, simple_loss=0.1378, pruned_loss=0.07808, over 1282.00 frames. ], tot_loss[loss=0.1084, simple_loss=0.1439, pruned_loss=0.0365, over 1978032.87 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:19:30,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 11:19:40,800 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 1.900e+02 2.307e+02 2.789e+02 7.825e+02, threshold=4.613e+02, percent-clipped=1.0 2022-12-08 11:19:53,680 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124628.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:02,945 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:24,759 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124663.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:50,164 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:50,238 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6784, 1.6022, 1.7450, 1.9433, 1.3754, 1.6713, 1.7369, 1.8367], device='cuda:1'), covar=tensor([0.0256, 0.0370, 0.0219, 0.0185, 0.0380, 0.0448, 0.0269, 0.0186], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0256, 0.0372, 0.0328, 0.0270, 0.0305, 0.0312, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:20:50,939 INFO [train.py:873] (1/4) Epoch 17, batch 3700, loss[loss=0.1355, simple_loss=0.1333, pruned_loss=0.06887, over 1227.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.144, pruned_loss=0.03695, over 1946582.43 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:20:56,534 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124699.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:00,445 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124704.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 11:21:07,859 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.026e+02 2.516e+02 3.153e+02 6.578e+02, threshold=5.031e+02, percent-clipped=7.0 2022-12-08 11:21:12,520 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:17,814 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124724.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:31,752 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:53,865 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 11:21:54,350 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:22:13,560 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:22:17,560 INFO [train.py:873] (1/4) Epoch 17, batch 3800, loss[loss=0.1165, simple_loss=0.151, pruned_loss=0.04099, over 13544.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1437, pruned_loss=0.03677, over 1959595.98 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 8.0 2022-12-08 11:22:35,562 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.150e+02 2.611e+02 3.247e+02 6.025e+02, threshold=5.221e+02, percent-clipped=4.0 2022-12-08 11:22:41,988 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:22:51,868 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 11:23:35,448 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:23:45,831 INFO [train.py:873] (1/4) Epoch 17, batch 3900, loss[loss=0.1151, simple_loss=0.1375, pruned_loss=0.04635, over 3899.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1436, pruned_loss=0.03671, over 1913153.02 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:23:46,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-08 11:23:52,407 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9043, 5.3714, 5.3450, 5.8349, 5.3503, 4.9333, 5.7973, 4.8410], device='cuda:1'), covar=tensor([0.0287, 0.0837, 0.0377, 0.0351, 0.0890, 0.0332, 0.0469, 0.0475], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0276, 0.0200, 0.0194, 0.0185, 0.0155, 0.0288, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 11:23:54,554 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 11:24:04,347 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.040e+02 2.475e+02 3.105e+02 8.016e+02, threshold=4.950e+02, percent-clipped=1.0 2022-12-08 11:24:17,221 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124928.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:24:28,273 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-08 11:24:59,255 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124976.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:25:13,840 INFO [train.py:873] (1/4) Epoch 17, batch 4000, loss[loss=0.103, simple_loss=0.1401, pruned_loss=0.03297, over 14539.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1433, pruned_loss=0.03597, over 1988490.09 frames. ], batch size: 49, lr: 4.62e-03, grad_scale: 8.0 2022-12-08 11:25:14,806 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124994.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:25:27,918 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125004.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 11:25:35,650 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3765, 1.0503, 1.2446, 0.8546, 1.1446, 1.3802, 1.1056, 1.1043], device='cuda:1'), covar=tensor([0.0561, 0.0912, 0.0719, 0.0437, 0.1063, 0.0864, 0.0531, 0.1332], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0173, 0.0142, 0.0128, 0.0144, 0.0156, 0.0135, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:25:36,307 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.040e+02 2.496e+02 2.950e+02 4.930e+02, threshold=4.992e+02, percent-clipped=0.0 2022-12-08 11:25:40,939 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125019.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:25:46,126 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9508, 2.5948, 5.0027, 3.3902, 4.8377, 2.2489, 3.7788, 4.6921], device='cuda:1'), covar=tensor([0.0561, 0.3480, 0.0388, 0.5509, 0.0490, 0.3477, 0.1292, 0.0598], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0200, 0.0217, 0.0270, 0.0235, 0.0205, 0.0202, 0.0218], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 11:26:09,623 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125052.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 11:26:24,865 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 11:26:39,445 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5539, 1.5462, 1.5183, 1.6988, 1.6251, 1.0173, 1.4622, 1.4170], device='cuda:1'), covar=tensor([0.0657, 0.0571, 0.0748, 0.0484, 0.0640, 0.0925, 0.0739, 0.0675], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0034, 0.0039, 0.0032, 0.0034, 0.0047, 0.0035, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 11:26:40,556 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:26:44,717 INFO [train.py:873] (1/4) Epoch 17, batch 4100, loss[loss=0.1374, simple_loss=0.1628, pruned_loss=0.05605, over 7770.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.1439, pruned_loss=0.03695, over 1880710.34 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:27:03,766 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.146e+02 2.604e+02 3.171e+02 6.787e+02, threshold=5.207e+02, percent-clipped=4.0 2022-12-08 11:27:15,552 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 11:27:15,930 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125128.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:27:22,533 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:27:40,874 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5003, 2.6934, 4.3944, 4.6001, 4.2629, 2.4743, 4.5716, 3.5356], device='cuda:1'), covar=tensor([0.0358, 0.1142, 0.0802, 0.0359, 0.0433, 0.1920, 0.0340, 0.0829], device='cuda:1'), in_proj_covar=tensor([0.0291, 0.0254, 0.0371, 0.0326, 0.0267, 0.0304, 0.0309, 0.0275], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:27:57,305 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:28:08,777 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125189.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:28:11,906 INFO [train.py:873] (1/4) Epoch 17, batch 4200, loss[loss=0.08966, simple_loss=0.133, pruned_loss=0.02316, over 13836.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1436, pruned_loss=0.03624, over 1987275.30 frames. ], batch size: 20, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:28:14,053 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8016, 1.3939, 1.7500, 1.2657, 1.5857, 1.8585, 1.5919, 1.5389], device='cuda:1'), covar=tensor([0.0868, 0.0712, 0.0646, 0.0858, 0.1325, 0.0804, 0.0672, 0.1829], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0174, 0.0142, 0.0127, 0.0144, 0.0156, 0.0136, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:28:31,908 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.000e+02 2.432e+02 3.049e+02 5.744e+02, threshold=4.863e+02, percent-clipped=3.0 2022-12-08 11:28:50,479 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2416, 1.4698, 4.1933, 2.0169, 4.0884, 4.2582, 3.5342, 4.6506], device='cuda:1'), covar=tensor([0.0232, 0.3252, 0.0348, 0.2050, 0.0342, 0.0372, 0.0556, 0.0145], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0158, 0.0166, 0.0166, 0.0178, 0.0132, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:29:24,844 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 11:29:40,187 INFO [train.py:873] (1/4) Epoch 17, batch 4300, loss[loss=0.09056, simple_loss=0.1364, pruned_loss=0.02237, over 14081.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1432, pruned_loss=0.0363, over 1892077.78 frames. ], batch size: 29, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:29:41,178 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125294.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:29:44,796 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5097, 1.4642, 1.5668, 1.6816, 1.6407, 1.0497, 1.3971, 1.4377], device='cuda:1'), covar=tensor([0.0993, 0.0891, 0.0666, 0.0600, 0.0538, 0.1031, 0.0756, 0.0843], device='cuda:1'), in_proj_covar=tensor([0.0035, 0.0035, 0.0039, 0.0032, 0.0034, 0.0047, 0.0035, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 11:29:58,839 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.107e+02 2.137e+02 2.626e+02 3.159e+02 7.655e+02, threshold=5.251e+02, percent-clipped=3.0 2022-12-08 11:30:02,804 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:08,118 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7752, 1.9899, 2.1367, 2.1762, 1.9787, 2.1360, 1.9114, 1.4853], device='cuda:1'), covar=tensor([0.1060, 0.1267, 0.0512, 0.0744, 0.1190, 0.0882, 0.1524, 0.2018], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0089, 0.0071, 0.0076, 0.0099, 0.0089, 0.0101, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:30:23,026 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:30,778 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 11:30:34,874 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:44,406 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125367.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:52,486 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:31:07,129 INFO [train.py:873] (1/4) Epoch 17, batch 4400, loss[loss=0.08962, simple_loss=0.1341, pruned_loss=0.02258, over 14338.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1427, pruned_loss=0.03538, over 1895950.23 frames. ], batch size: 39, lr: 4.61e-03, grad_scale: 8.0 2022-12-08 11:31:26,863 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.784e+01 2.190e+02 2.587e+02 3.207e+02 7.157e+02, threshold=5.175e+02, percent-clipped=2.0 2022-12-08 11:31:27,839 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1554, 3.9739, 3.8772, 4.2586, 3.8178, 3.5488, 4.2451, 4.0201], device='cuda:1'), covar=tensor([0.0690, 0.1017, 0.0892, 0.0640, 0.1007, 0.0792, 0.0638, 0.0850], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0147, 0.0150, 0.0165, 0.0151, 0.0126, 0.0173, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:31:28,771 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125417.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:31:32,392 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125421.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:31:46,353 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:19,965 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125476.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:25,338 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125482.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:26,965 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125484.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:28,154 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.53 vs. limit=5.0 2022-12-08 11:32:35,101 INFO [train.py:873] (1/4) Epoch 17, batch 4500, loss[loss=0.1018, simple_loss=0.1396, pruned_loss=0.03205, over 6923.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.1428, pruned_loss=0.03579, over 1867175.51 frames. ], batch size: 100, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:32:42,753 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 11:32:54,648 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 2.002e+02 2.492e+02 3.086e+02 6.444e+02, threshold=4.984e+02, percent-clipped=2.0 2022-12-08 11:33:01,480 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:34:01,914 INFO [train.py:873] (1/4) Epoch 17, batch 4600, loss[loss=0.1041, simple_loss=0.1448, pruned_loss=0.03173, over 14441.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.1432, pruned_loss=0.03557, over 1954819.10 frames. ], batch size: 53, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:34:11,411 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:34:22,483 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.136e+02 2.137e+02 2.692e+02 3.129e+02 4.972e+02, threshold=5.384e+02, percent-clipped=0.0 2022-12-08 11:34:42,277 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 11:34:54,259 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 11:35:05,293 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125664.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:35:30,460 INFO [train.py:873] (1/4) Epoch 17, batch 4700, loss[loss=0.1086, simple_loss=0.1497, pruned_loss=0.03379, over 14242.00 frames. ], tot_loss[loss=0.1084, simple_loss=0.144, pruned_loss=0.03641, over 1959322.14 frames. ], batch size: 35, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:35:47,174 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125712.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:35:47,982 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6873, 3.4724, 3.3280, 3.6883, 3.4912, 3.6721, 3.7654, 3.1304], device='cuda:1'), covar=tensor([0.0447, 0.0985, 0.0543, 0.0503, 0.0751, 0.0384, 0.0512, 0.0621], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0278, 0.0201, 0.0195, 0.0185, 0.0158, 0.0288, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 11:35:50,739 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.065e+02 2.672e+02 3.737e+02 1.296e+03, threshold=5.344e+02, percent-clipped=9.0 2022-12-08 11:35:57,954 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125725.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:04,509 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:44,455 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:50,817 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:52,582 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125786.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:54,062 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9600, 1.2852, 2.0234, 1.3230, 1.9989, 2.0941, 1.6922, 2.1647], device='cuda:1'), covar=tensor([0.0339, 0.2217, 0.0542, 0.1820, 0.0593, 0.0573, 0.1153, 0.0393], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0158, 0.0166, 0.0166, 0.0178, 0.0131, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:36:58,259 INFO [train.py:873] (1/4) Epoch 17, batch 4800, loss[loss=0.1172, simple_loss=0.1451, pruned_loss=0.04463, over 5998.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1435, pruned_loss=0.03615, over 1950319.49 frames. ], batch size: 100, lr: 4.61e-03, grad_scale: 8.0 2022-12-08 11:37:19,643 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 1.900e+02 2.405e+02 3.076e+02 6.689e+02, threshold=4.810e+02, percent-clipped=2.0 2022-12-08 11:37:32,557 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125832.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:38:26,384 INFO [train.py:873] (1/4) Epoch 17, batch 4900, loss[loss=0.08679, simple_loss=0.126, pruned_loss=0.02378, over 14057.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.1444, pruned_loss=0.03672, over 1975884.75 frames. ], batch size: 22, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:38:40,522 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4014, 2.3386, 2.1235, 2.3539, 2.0265, 1.5487, 2.0662, 2.4700], device='cuda:1'), covar=tensor([0.1041, 0.0553, 0.1091, 0.1675, 0.1869, 0.0898, 0.1421, 0.0672], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0035, 0.0040, 0.0033, 0.0034, 0.0048, 0.0036, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 11:38:47,414 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.229e+02 2.578e+02 3.069e+02 1.052e+03, threshold=5.155e+02, percent-clipped=5.0 2022-12-08 11:39:24,581 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125959.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:39:54,283 INFO [train.py:873] (1/4) Epoch 17, batch 5000, loss[loss=0.1068, simple_loss=0.1437, pruned_loss=0.03498, over 13521.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1435, pruned_loss=0.03612, over 1959498.43 frames. ], batch size: 100, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:39:58,741 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:11,510 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126012.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 11:40:15,916 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.233e+02 2.614e+02 3.180e+02 5.706e+02, threshold=5.227e+02, percent-clipped=5.0 2022-12-08 11:40:28,947 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:52,818 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126059.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:53,523 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126060.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:08,274 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126077.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:11,115 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:12,047 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126081.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:22,061 INFO [train.py:873] (1/4) Epoch 17, batch 5100, loss[loss=0.1269, simple_loss=0.1522, pruned_loss=0.05083, over 14241.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.1431, pruned_loss=0.03562, over 1989007.42 frames. ], batch size: 46, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:41:34,638 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2022-12-08 11:41:42,935 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.125e+02 2.682e+02 3.243e+02 7.763e+02, threshold=5.364e+02, percent-clipped=5.0 2022-12-08 11:41:50,078 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126125.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:42:15,365 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8211, 0.7324, 0.7584, 0.6882, 0.7349, 0.6457, 0.5863, 0.7248], device='cuda:1'), covar=tensor([0.0167, 0.0178, 0.0161, 0.0172, 0.0181, 0.0330, 0.0202, 0.0284], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0021, 0.0019, 0.0021, 0.0020, 0.0033, 0.0027, 0.0031], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 11:42:17,080 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3169, 1.3549, 2.4855, 1.4340, 2.4273, 2.4557, 1.9280, 2.5953], device='cuda:1'), covar=tensor([0.0315, 0.2566, 0.0435, 0.1944, 0.0566, 0.0572, 0.1053, 0.0345], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0156, 0.0159, 0.0166, 0.0167, 0.0179, 0.0132, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:42:42,312 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 11:42:49,533 INFO [train.py:873] (1/4) Epoch 17, batch 5200, loss[loss=0.1363, simple_loss=0.1336, pruned_loss=0.06955, over 1217.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1433, pruned_loss=0.03567, over 2029146.25 frames. ], batch size: 100, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:42:54,794 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1046, 4.6565, 4.5147, 5.0593, 4.7955, 4.4317, 5.0188, 4.2380], device='cuda:1'), covar=tensor([0.0318, 0.0979, 0.0410, 0.0385, 0.0725, 0.0517, 0.0515, 0.0481], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0276, 0.0199, 0.0194, 0.0183, 0.0157, 0.0286, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 11:43:10,801 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.118e+02 2.718e+02 3.349e+02 1.527e+03, threshold=5.437e+02, percent-clipped=4.0 2022-12-08 11:43:19,865 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9427, 1.5950, 3.1097, 2.7603, 3.0137, 3.1439, 2.3456, 3.0755], device='cuda:1'), covar=tensor([0.1317, 0.1490, 0.0174, 0.0457, 0.0344, 0.0226, 0.0569, 0.0206], device='cuda:1'), in_proj_covar=tensor([0.0144, 0.0157, 0.0130, 0.0167, 0.0147, 0.0142, 0.0124, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 11:43:48,200 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:44:17,889 INFO [train.py:873] (1/4) Epoch 17, batch 5300, loss[loss=0.1474, simple_loss=0.1411, pruned_loss=0.07683, over 1202.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.1433, pruned_loss=0.03621, over 1904623.65 frames. ], batch size: 100, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:44:23,584 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8514, 4.0031, 4.1853, 3.7615, 4.0471, 4.1747, 1.6461, 3.8070], device='cuda:1'), covar=tensor([0.0365, 0.0335, 0.0343, 0.0457, 0.0347, 0.0306, 0.3084, 0.0312], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0174, 0.0146, 0.0146, 0.0206, 0.0142, 0.0156, 0.0193], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 11:44:30,690 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:44:39,315 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 2.097e+02 2.584e+02 3.197e+02 6.466e+02, threshold=5.169e+02, percent-clipped=5.0 2022-12-08 11:45:12,375 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126354.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:19,165 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4268, 2.7199, 2.7077, 2.8212, 2.2780, 2.8452, 2.6300, 1.6422], device='cuda:1'), covar=tensor([0.1004, 0.0721, 0.0664, 0.0529, 0.0915, 0.0560, 0.0910, 0.1790], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0089, 0.0070, 0.0076, 0.0099, 0.0089, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 11:45:34,493 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.55 vs. limit=5.0 2022-12-08 11:45:35,440 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:36,308 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126381.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:46,520 INFO [train.py:873] (1/4) Epoch 17, batch 5400, loss[loss=0.11, simple_loss=0.1425, pruned_loss=0.03882, over 14148.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1427, pruned_loss=0.03561, over 1905249.19 frames. ], batch size: 84, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:46:08,053 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.193e+02 2.737e+02 3.423e+02 8.531e+02, threshold=5.474e+02, percent-clipped=2.0 2022-12-08 11:46:19,135 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:46:29,888 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:46:37,519 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:47:12,455 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.0656, 5.5610, 5.5710, 6.0563, 5.6541, 4.8951, 6.0242, 4.9035], device='cuda:1'), covar=tensor([0.0350, 0.0910, 0.0312, 0.0365, 0.0690, 0.0395, 0.0454, 0.0510], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0277, 0.0201, 0.0195, 0.0185, 0.0158, 0.0287, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 11:47:14,891 INFO [train.py:873] (1/4) Epoch 17, batch 5500, loss[loss=0.1274, simple_loss=0.1186, pruned_loss=0.06813, over 1289.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1425, pruned_loss=0.03565, over 1920702.04 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:47:31,257 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126511.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:47:36,283 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.086e+02 2.565e+02 2.972e+02 5.597e+02, threshold=5.130e+02, percent-clipped=1.0 2022-12-08 11:48:42,923 INFO [train.py:873] (1/4) Epoch 17, batch 5600, loss[loss=0.0813, simple_loss=0.1287, pruned_loss=0.01696, over 14307.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1434, pruned_loss=0.03639, over 1915857.84 frames. ], batch size: 39, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:49:04,887 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 1.973e+02 2.464e+02 3.253e+02 5.344e+02, threshold=4.927e+02, percent-clipped=2.0 2022-12-08 11:49:36,605 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126654.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:50:10,195 INFO [train.py:873] (1/4) Epoch 17, batch 5700, loss[loss=0.1047, simple_loss=0.1491, pruned_loss=0.03016, over 14506.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1432, pruned_loss=0.03633, over 1954070.39 frames. ], batch size: 49, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:50:18,647 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126702.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:50:28,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5149, 3.2927, 3.2138, 3.4904, 3.3404, 3.4846, 3.5429, 2.9445], device='cuda:1'), covar=tensor([0.0541, 0.0915, 0.0502, 0.0514, 0.0749, 0.0385, 0.0571, 0.0650], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0275, 0.0199, 0.0195, 0.0185, 0.0157, 0.0287, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 11:50:32,276 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.662e+01 2.224e+02 2.546e+02 3.152e+02 6.937e+02, threshold=5.092e+02, percent-clipped=3.0 2022-12-08 11:50:35,999 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5628, 1.5405, 4.3001, 1.9407, 4.2304, 4.6613, 4.1948, 4.9723], device='cuda:1'), covar=tensor([0.0299, 0.3363, 0.0433, 0.2393, 0.0382, 0.0429, 0.0369, 0.0232], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0154, 0.0158, 0.0165, 0.0165, 0.0177, 0.0130, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:50:48,051 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:51:30,281 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6948, 2.9418, 4.5570, 3.2585, 4.4951, 4.3159, 4.2998, 3.9610], device='cuda:1'), covar=tensor([0.0922, 0.3022, 0.0915, 0.1863, 0.0695, 0.0970, 0.1441, 0.1688], device='cuda:1'), in_proj_covar=tensor([0.0356, 0.0313, 0.0397, 0.0300, 0.0368, 0.0326, 0.0365, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:51:37,724 INFO [train.py:873] (1/4) Epoch 17, batch 5800, loss[loss=0.1133, simple_loss=0.1486, pruned_loss=0.03903, over 14363.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1434, pruned_loss=0.03595, over 2001438.63 frames. ], batch size: 41, lr: 4.59e-03, grad_scale: 4.0 2022-12-08 11:51:49,396 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:51:58,017 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126816.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:52:01,106 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.070e+02 2.153e+02 2.501e+02 2.856e+02 4.579e+02, threshold=5.001e+02, percent-clipped=0.0 2022-12-08 11:52:51,090 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126877.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:53:04,763 INFO [train.py:873] (1/4) Epoch 17, batch 5900, loss[loss=0.09501, simple_loss=0.1273, pruned_loss=0.03135, over 6915.00 frames. ], tot_loss[loss=0.107, simple_loss=0.1427, pruned_loss=0.03561, over 1985662.19 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 2.0 2022-12-08 11:53:10,507 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1601, 3.2436, 3.3795, 3.2991, 3.3065, 2.8910, 1.4255, 3.1314], device='cuda:1'), covar=tensor([0.0443, 0.0417, 0.0405, 0.0353, 0.0334, 0.0778, 0.3015, 0.0325], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0174, 0.0145, 0.0146, 0.0205, 0.0141, 0.0156, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 11:53:29,086 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.272e+02 2.688e+02 3.245e+02 5.334e+02, threshold=5.377e+02, percent-clipped=3.0 2022-12-08 11:53:57,328 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 11:54:32,807 INFO [train.py:873] (1/4) Epoch 17, batch 6000, loss[loss=0.09126, simple_loss=0.1092, pruned_loss=0.03667, over 2650.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1435, pruned_loss=0.0364, over 1930241.54 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:54:32,807 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 11:54:41,593 INFO [train.py:905] (1/4) Epoch 17, validation: loss=0.1381, simple_loss=0.176, pruned_loss=0.05009, over 857387.00 frames. 2022-12-08 11:54:41,594 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 11:55:05,926 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 1.999e+02 2.582e+02 3.270e+02 6.669e+02, threshold=5.164e+02, percent-clipped=4.0 2022-12-08 11:55:20,276 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:55:41,917 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7221, 5.6124, 5.1343, 5.8709, 5.2803, 5.2110, 5.8913, 5.5578], device='cuda:1'), covar=tensor([0.0643, 0.0677, 0.0816, 0.0443, 0.0634, 0.0345, 0.0525, 0.0672], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0145, 0.0148, 0.0163, 0.0150, 0.0125, 0.0171, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:55:43,848 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1138, 2.7699, 3.6720, 2.5840, 2.2774, 3.2469, 1.7354, 3.2244], device='cuda:1'), covar=tensor([0.0939, 0.1337, 0.0591, 0.1622, 0.2055, 0.0824, 0.3020, 0.0794], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0095, 0.0100, 0.0115, 0.0091, 0.0117, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 11:56:01,545 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:56:09,389 INFO [train.py:873] (1/4) Epoch 17, batch 6100, loss[loss=0.1514, simple_loss=0.142, pruned_loss=0.08044, over 1243.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.143, pruned_loss=0.03611, over 1939793.74 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:56:20,845 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:56:28,153 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5351, 2.5307, 2.0337, 2.6545, 2.4634, 2.4724, 2.2804, 2.1481], device='cuda:1'), covar=tensor([0.1099, 0.1313, 0.2762, 0.0824, 0.1313, 0.1122, 0.1499, 0.2084], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0287, 0.0259, 0.0285, 0.0323, 0.0301, 0.0252, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:56:33,263 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.062e+02 2.398e+02 3.031e+02 5.398e+02, threshold=4.796e+02, percent-clipped=2.0 2022-12-08 11:56:35,642 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 11:57:02,818 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127154.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:57:18,875 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:57:37,193 INFO [train.py:873] (1/4) Epoch 17, batch 6200, loss[loss=0.1056, simple_loss=0.1519, pruned_loss=0.02962, over 14023.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1427, pruned_loss=0.03505, over 1952243.66 frames. ], batch size: 22, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:57:48,199 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7187, 3.8561, 3.9925, 3.6153, 3.8892, 3.8562, 1.6200, 3.6954], device='cuda:1'), covar=tensor([0.0352, 0.0372, 0.0346, 0.0455, 0.0335, 0.0421, 0.2957, 0.0304], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0174, 0.0145, 0.0147, 0.0206, 0.0142, 0.0157, 0.0192], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 11:58:01,694 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.117e+02 2.626e+02 3.289e+02 8.585e+02, threshold=5.253e+02, percent-clipped=4.0 2022-12-08 11:58:27,567 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127249.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:58:27,770 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 11:58:58,366 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 11:59:05,810 INFO [train.py:873] (1/4) Epoch 17, batch 6300, loss[loss=0.1107, simple_loss=0.1476, pruned_loss=0.03689, over 13524.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.1421, pruned_loss=0.03459, over 1954413.25 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:59:09,146 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-08 11:59:10,084 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2022-12-08 11:59:21,213 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127310.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:59:29,649 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.074e+02 2.479e+02 3.103e+02 6.398e+02, threshold=4.958e+02, percent-clipped=2.0 2022-12-08 11:59:30,702 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2231, 4.0368, 3.9429, 4.2684, 3.8995, 3.5313, 4.3164, 4.1317], device='cuda:1'), covar=tensor([0.0709, 0.0909, 0.0889, 0.0606, 0.0880, 0.0820, 0.0624, 0.0815], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0144, 0.0147, 0.0163, 0.0149, 0.0124, 0.0169, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:59:35,866 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2022-12-08 11:59:36,300 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8948, 2.4146, 3.7556, 2.6853, 3.7400, 3.5482, 3.5820, 3.1509], device='cuda:1'), covar=tensor([0.0968, 0.3190, 0.1022, 0.2136, 0.0908, 0.1176, 0.1510, 0.1727], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0311, 0.0393, 0.0300, 0.0367, 0.0322, 0.0363, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 11:59:37,066 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0151, 2.1016, 1.9442, 2.1330, 1.8066, 1.9310, 2.1035, 2.0308], device='cuda:1'), covar=tensor([0.1118, 0.1158, 0.1197, 0.1009, 0.1613, 0.0929, 0.1067, 0.0989], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0144, 0.0147, 0.0163, 0.0149, 0.0124, 0.0169, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 11:59:56,644 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8554, 1.2725, 1.9956, 1.2500, 1.9331, 2.0573, 1.7348, 2.1543], device='cuda:1'), covar=tensor([0.0317, 0.2272, 0.0567, 0.2116, 0.0612, 0.0623, 0.1093, 0.0424], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0154, 0.0157, 0.0165, 0.0165, 0.0178, 0.0130, 0.0149], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:00:33,560 INFO [train.py:873] (1/4) Epoch 17, batch 6400, loss[loss=0.1028, simple_loss=0.1246, pruned_loss=0.04045, over 3870.00 frames. ], tot_loss[loss=0.1068, simple_loss=0.1427, pruned_loss=0.03546, over 1950812.56 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 8.0 2022-12-08 12:00:42,388 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0885, 1.8716, 2.1201, 2.0132, 1.8544, 1.9163, 1.7021, 1.2845], device='cuda:1'), covar=tensor([0.0261, 0.0498, 0.0230, 0.0247, 0.0341, 0.0352, 0.0445, 0.0688], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:00:58,034 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.677e+01 2.186e+02 2.674e+02 3.273e+02 5.591e+02, threshold=5.347e+02, percent-clipped=3.0 2022-12-08 12:01:04,719 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127427.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:01:43,616 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:01:57,909 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127488.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:02:01,913 INFO [train.py:873] (1/4) Epoch 17, batch 6500, loss[loss=0.1043, simple_loss=0.127, pruned_loss=0.04084, over 3865.00 frames. ], tot_loss[loss=0.107, simple_loss=0.143, pruned_loss=0.03552, over 1978308.04 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 8.0 2022-12-08 12:02:02,097 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1720, 3.1802, 4.0262, 2.8305, 2.6390, 3.4620, 2.0241, 3.5700], device='cuda:1'), covar=tensor([0.1120, 0.1004, 0.0441, 0.2110, 0.1651, 0.0928, 0.2580, 0.0748], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0103, 0.0097, 0.0101, 0.0116, 0.0091, 0.0118, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:02:25,881 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.139e+02 2.589e+02 3.452e+02 6.629e+02, threshold=5.179e+02, percent-clipped=2.0 2022-12-08 12:02:25,982 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:02:34,135 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:27,605 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127590.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:29,961 INFO [train.py:873] (1/4) Epoch 17, batch 6600, loss[loss=0.1134, simple_loss=0.1366, pruned_loss=0.04512, over 3852.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1429, pruned_loss=0.03587, over 1985367.47 frames. ], batch size: 100, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:03:40,629 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:54,225 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.073e+02 2.578e+02 3.202e+02 5.753e+02, threshold=5.156e+02, percent-clipped=2.0 2022-12-08 12:03:57,790 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7861, 1.6559, 1.8345, 1.9830, 1.4554, 1.7081, 1.8036, 1.8856], device='cuda:1'), covar=tensor([0.0267, 0.0385, 0.0219, 0.0219, 0.0384, 0.0420, 0.0344, 0.0262], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0258, 0.0375, 0.0330, 0.0270, 0.0306, 0.0310, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:04:30,686 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127662.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:04:47,356 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9015, 1.3086, 2.0082, 1.2495, 1.9766, 2.0583, 1.7808, 2.1548], device='cuda:1'), covar=tensor([0.0350, 0.2086, 0.0541, 0.2052, 0.0608, 0.0572, 0.1013, 0.0384], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0153, 0.0158, 0.0165, 0.0166, 0.0177, 0.0131, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:04:57,442 INFO [train.py:873] (1/4) Epoch 17, batch 6700, loss[loss=0.1285, simple_loss=0.1628, pruned_loss=0.04709, over 14469.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1434, pruned_loss=0.03558, over 2014813.05 frames. ], batch size: 51, lr: 4.57e-03, grad_scale: 4.0 2022-12-08 12:05:21,963 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.055e+02 2.568e+02 3.169e+02 6.326e+02, threshold=5.135e+02, percent-clipped=1.0 2022-12-08 12:05:23,966 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127723.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:05:40,440 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9682, 1.4266, 3.0726, 1.6340, 3.2589, 3.1355, 2.3571, 3.3425], device='cuda:1'), covar=tensor([0.0324, 0.2980, 0.0469, 0.2053, 0.0327, 0.0419, 0.0949, 0.0232], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0153, 0.0158, 0.0165, 0.0166, 0.0177, 0.0131, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:05:46,948 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-08 12:06:05,565 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.53 vs. limit=2.0 2022-12-08 12:06:07,613 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8893, 0.7862, 0.8400, 0.7799, 0.8400, 0.6720, 0.7019, 0.8239], device='cuda:1'), covar=tensor([0.0162, 0.0170, 0.0146, 0.0173, 0.0160, 0.0262, 0.0204, 0.0240], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:06:13,129 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127779.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:06:16,541 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127783.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:06:25,290 INFO [train.py:873] (1/4) Epoch 17, batch 6800, loss[loss=0.1155, simple_loss=0.1464, pruned_loss=0.04228, over 4942.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1436, pruned_loss=0.03547, over 2038205.67 frames. ], batch size: 100, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:06:50,326 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.046e+02 2.510e+02 3.119e+02 5.930e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 12:07:07,273 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127840.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:16,154 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:35,242 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:46,423 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127885.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:53,422 INFO [train.py:873] (1/4) Epoch 17, batch 6900, loss[loss=0.1052, simple_loss=0.147, pruned_loss=0.03168, over 14571.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1431, pruned_loss=0.03541, over 1977521.22 frames. ], batch size: 34, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:08:03,984 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127905.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:09,350 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:17,921 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.924e+01 2.068e+02 2.418e+02 3.139e+02 1.064e+03, threshold=4.836e+02, percent-clipped=7.0 2022-12-08 12:08:28,439 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:46,261 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127953.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:09:19,918 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 12:09:20,963 INFO [train.py:873] (1/4) Epoch 17, batch 7000, loss[loss=0.0981, simple_loss=0.1426, pruned_loss=0.02681, over 14451.00 frames. ], tot_loss[loss=0.1068, simple_loss=0.1431, pruned_loss=0.03523, over 2011462.34 frames. ], batch size: 51, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:09:23,717 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4536, 1.3834, 3.5505, 1.6132, 3.3176, 3.5938, 2.4758, 3.7964], device='cuda:1'), covar=tensor([0.0249, 0.3105, 0.0369, 0.2262, 0.0764, 0.0380, 0.0851, 0.0198], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0153, 0.0158, 0.0166, 0.0165, 0.0177, 0.0131, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:09:43,962 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:09:44,946 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1333, 2.7049, 2.8685, 1.8904, 2.5612, 2.8061, 3.1181, 2.5450], device='cuda:1'), covar=tensor([0.0657, 0.0876, 0.0869, 0.1444, 0.1157, 0.0647, 0.0624, 0.1304], device='cuda:1'), in_proj_covar=tensor([0.0151, 0.0171, 0.0141, 0.0126, 0.0144, 0.0157, 0.0136, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 12:09:47,264 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.259e+02 2.605e+02 3.114e+02 4.629e+02, threshold=5.210e+02, percent-clipped=0.0 2022-12-08 12:10:02,234 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-08 12:10:18,314 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2530, 1.6521, 2.4462, 1.9950, 2.3190, 1.6563, 2.0024, 2.2598], device='cuda:1'), covar=tensor([0.1971, 0.3221, 0.0786, 0.1694, 0.1205, 0.2293, 0.1020, 0.1035], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0200, 0.0221, 0.0269, 0.0238, 0.0202, 0.0203, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 12:10:21,458 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2022-12-08 12:10:41,157 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128083.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:10:49,317 INFO [train.py:873] (1/4) Epoch 17, batch 7100, loss[loss=0.1126, simple_loss=0.1201, pruned_loss=0.05252, over 2547.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1433, pruned_loss=0.03564, over 2011568.39 frames. ], batch size: 100, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:10:58,679 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 12:11:14,254 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.219e+02 2.800e+02 3.533e+02 6.463e+02, threshold=5.601e+02, percent-clipped=4.0 2022-12-08 12:11:22,292 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128131.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:11:25,688 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128135.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:11:46,412 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 12:12:09,566 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128185.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:16,537 INFO [train.py:873] (1/4) Epoch 17, batch 7200, loss[loss=0.09767, simple_loss=0.125, pruned_loss=0.03515, over 3902.00 frames. ], tot_loss[loss=0.1074, simple_loss=0.1432, pruned_loss=0.0358, over 1947930.18 frames. ], batch size: 100, lr: 4.56e-03, grad_scale: 8.0 2022-12-08 12:12:28,378 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:42,993 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.064e+02 2.461e+02 3.111e+02 6.198e+02, threshold=4.923e+02, percent-clipped=2.0 2022-12-08 12:12:47,594 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:52,151 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:54,992 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2022-12-08 12:13:01,105 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 12:13:44,321 INFO [train.py:873] (1/4) Epoch 17, batch 7300, loss[loss=0.1115, simple_loss=0.1471, pruned_loss=0.03793, over 14255.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1426, pruned_loss=0.0356, over 1937186.94 frames. ], batch size: 57, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:14:05,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8564, 0.8736, 0.6842, 0.8799, 0.9016, 0.4024, 0.8015, 0.8527], device='cuda:1'), covar=tensor([0.0426, 0.0488, 0.0592, 0.0454, 0.0351, 0.0298, 0.0762, 0.0714], device='cuda:1'), in_proj_covar=tensor([0.0037, 0.0036, 0.0040, 0.0033, 0.0035, 0.0048, 0.0037, 0.0039], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:14:06,060 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:14:08,085 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128320.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:14:10,509 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.143e+02 2.591e+02 3.352e+02 1.032e+03, threshold=5.183e+02, percent-clipped=4.0 2022-12-08 12:14:23,572 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3448, 2.2017, 2.5466, 1.7093, 1.7542, 2.3229, 1.4359, 2.3129], device='cuda:1'), covar=tensor([0.1294, 0.1259, 0.0864, 0.2184, 0.2224, 0.0969, 0.3187, 0.1032], device='cuda:1'), in_proj_covar=tensor([0.0085, 0.0101, 0.0095, 0.0099, 0.0113, 0.0090, 0.0116, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:14:45,558 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7594, 1.9168, 2.6041, 2.1739, 2.6383, 2.5773, 2.4340, 2.4366], device='cuda:1'), covar=tensor([0.0881, 0.2833, 0.0914, 0.1826, 0.0707, 0.1138, 0.0923, 0.1341], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0311, 0.0393, 0.0299, 0.0365, 0.0323, 0.0362, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:14:47,736 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128366.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:14:54,124 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6978, 2.3254, 3.5012, 2.6995, 3.6080, 3.4149, 3.3276, 2.9328], device='cuda:1'), covar=tensor([0.0982, 0.3227, 0.1282, 0.2116, 0.0826, 0.1208, 0.1430, 0.1949], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0311, 0.0394, 0.0299, 0.0365, 0.0323, 0.0362, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:15:00,840 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128381.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:15:03,356 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4084, 1.7896, 2.3003, 1.9540, 2.4296, 2.2574, 2.1998, 2.2563], device='cuda:1'), covar=tensor([0.0633, 0.2376, 0.0823, 0.1327, 0.0681, 0.1110, 0.0739, 0.1068], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0311, 0.0393, 0.0299, 0.0365, 0.0323, 0.0362, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:15:11,325 INFO [train.py:873] (1/4) Epoch 17, batch 7400, loss[loss=0.09913, simple_loss=0.1422, pruned_loss=0.02803, over 14439.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1424, pruned_loss=0.03515, over 1968728.31 frames. ], batch size: 53, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:15:29,166 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128413.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:15:37,701 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.104e+02 2.629e+02 3.153e+02 1.005e+03, threshold=5.259e+02, percent-clipped=2.0 2022-12-08 12:15:42,877 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8639, 1.9352, 2.0678, 1.4248, 1.5417, 1.8989, 1.2474, 1.9392], device='cuda:1'), covar=tensor([0.1474, 0.2202, 0.1017, 0.2756, 0.2879, 0.1196, 0.3359, 0.1126], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0102, 0.0095, 0.0100, 0.0114, 0.0091, 0.0117, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:15:48,005 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:10,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0117, 2.1105, 1.9250, 2.1354, 1.7846, 2.0040, 2.0886, 2.0230], device='cuda:1'), covar=tensor([0.0979, 0.1200, 0.1153, 0.0873, 0.1377, 0.0873, 0.1090, 0.1123], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0147, 0.0150, 0.0165, 0.0151, 0.0126, 0.0173, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:16:22,021 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:29,521 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:38,247 INFO [train.py:873] (1/4) Epoch 17, batch 7500, loss[loss=0.09086, simple_loss=0.1321, pruned_loss=0.02479, over 14286.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1434, pruned_loss=0.03622, over 1988017.59 frames. ], batch size: 31, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:16:39,599 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2022-12-08 12:16:49,195 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:17:04,009 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.021e+02 2.073e+02 2.518e+02 3.139e+02 6.346e+02, threshold=5.035e+02, percent-clipped=2.0 2022-12-08 12:17:08,311 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:17:08,329 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1624, 1.4276, 1.7216, 1.6817, 1.5721, 1.6428, 1.4210, 1.2896], device='cuda:1'), covar=tensor([0.1400, 0.1464, 0.0612, 0.0652, 0.1439, 0.1113, 0.1566, 0.1770], device='cuda:1'), in_proj_covar=tensor([0.0138, 0.0090, 0.0070, 0.0076, 0.0099, 0.0090, 0.0101, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 12:18:06,085 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128554.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:18:06,901 INFO [train.py:873] (1/4) Epoch 18, batch 0, loss[loss=0.1271, simple_loss=0.1751, pruned_loss=0.03957, over 14324.00 frames. ], tot_loss[loss=0.1271, simple_loss=0.1751, pruned_loss=0.03957, over 14324.00 frames. ], batch size: 55, lr: 4.43e-03, grad_scale: 8.0 2022-12-08 12:18:06,901 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 12:18:14,462 INFO [train.py:905] (1/4) Epoch 18, validation: loss=0.1457, simple_loss=0.1856, pruned_loss=0.05295, over 857387.00 frames. 2022-12-08 12:18:14,462 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 12:18:23,691 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6682, 3.4312, 3.2364, 3.3817, 3.5801, 3.6034, 3.6456, 3.6646], device='cuda:1'), covar=tensor([0.0929, 0.0667, 0.1997, 0.2501, 0.0816, 0.0906, 0.1012, 0.0785], device='cuda:1'), in_proj_covar=tensor([0.0393, 0.0277, 0.0454, 0.0569, 0.0355, 0.0455, 0.0392, 0.0400], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:18:33,523 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:18:54,868 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8112, 3.5372, 2.6895, 3.9249, 3.8822, 3.8498, 3.3162, 2.8285], device='cuda:1'), covar=tensor([0.0848, 0.1137, 0.3173, 0.0646, 0.0797, 0.1084, 0.1202, 0.2858], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0287, 0.0258, 0.0286, 0.0321, 0.0302, 0.0251, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:19:16,502 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 5.261e+01 1.855e+02 2.639e+02 3.708e+02 1.096e+03, threshold=5.279e+02, percent-clipped=10.0 2022-12-08 12:19:24,580 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8360, 4.4782, 4.3051, 4.8508, 4.5370, 4.3245, 4.8036, 4.0545], device='cuda:1'), covar=tensor([0.0396, 0.1002, 0.0429, 0.0412, 0.0751, 0.0611, 0.0555, 0.0547], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0279, 0.0203, 0.0195, 0.0186, 0.0159, 0.0289, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:19:43,876 INFO [train.py:873] (1/4) Epoch 18, batch 100, loss[loss=0.08359, simple_loss=0.1268, pruned_loss=0.02019, over 14095.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.142, pruned_loss=0.03388, over 855555.76 frames. ], batch size: 22, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:19:50,204 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4153, 1.3514, 1.3594, 1.5812, 1.5337, 1.0153, 1.1833, 1.3836], device='cuda:1'), covar=tensor([0.0793, 0.0739, 0.0651, 0.0481, 0.0583, 0.0860, 0.0943, 0.0654], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0035, 0.0040, 0.0033, 0.0034, 0.0048, 0.0036, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:20:01,879 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128676.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:20:42,730 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.050e+02 2.470e+02 3.144e+02 6.150e+02, threshold=4.940e+02, percent-clipped=2.0 2022-12-08 12:21:09,630 INFO [train.py:873] (1/4) Epoch 18, batch 200, loss[loss=0.1549, simple_loss=0.1756, pruned_loss=0.0671, over 8626.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.142, pruned_loss=0.03462, over 1312450.84 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:21:18,020 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8410, 1.8587, 1.6506, 1.8767, 1.7680, 1.8307, 1.8062, 1.6776], device='cuda:1'), covar=tensor([0.1113, 0.0883, 0.1826, 0.0836, 0.0964, 0.0700, 0.1460, 0.1158], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0289, 0.0259, 0.0287, 0.0322, 0.0304, 0.0252, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:21:22,421 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128769.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:21:32,315 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:21:54,960 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2615, 2.2396, 3.1628, 2.4658, 3.2051, 3.1100, 3.0119, 2.6724], device='cuda:1'), covar=tensor([0.0952, 0.2924, 0.1065, 0.1830, 0.0788, 0.1081, 0.1300, 0.1970], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0311, 0.0391, 0.0296, 0.0364, 0.0322, 0.0360, 0.0297], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:22:09,532 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 1.948e+02 2.392e+02 2.813e+02 5.621e+02, threshold=4.784e+02, percent-clipped=1.0 2022-12-08 12:22:25,632 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128842.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:22:37,492 INFO [train.py:873] (1/4) Epoch 18, batch 300, loss[loss=0.1195, simple_loss=0.136, pruned_loss=0.05148, over 3799.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1416, pruned_loss=0.03473, over 1554902.54 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:22:59,041 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5646, 5.2406, 4.9063, 5.1573, 5.1142, 5.4342, 5.4676, 5.4900], device='cuda:1'), covar=tensor([0.0548, 0.0363, 0.1561, 0.1972, 0.0609, 0.0630, 0.0641, 0.0680], device='cuda:1'), in_proj_covar=tensor([0.0392, 0.0278, 0.0453, 0.0567, 0.0351, 0.0454, 0.0392, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:23:08,225 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128890.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:23:17,888 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0789, 3.1689, 3.2544, 3.1109, 3.1663, 3.0002, 1.4081, 2.9576], device='cuda:1'), covar=tensor([0.0475, 0.0430, 0.0356, 0.0409, 0.0361, 0.0817, 0.2987, 0.0321], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0177, 0.0145, 0.0149, 0.0208, 0.0144, 0.0158, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:23:37,894 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.277e+02 2.735e+02 3.513e+02 6.767e+02, threshold=5.469e+02, percent-clipped=7.0 2022-12-08 12:23:48,964 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 12:23:54,977 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 12:24:02,506 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128951.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:24:05,631 INFO [train.py:873] (1/4) Epoch 18, batch 400, loss[loss=0.1006, simple_loss=0.1394, pruned_loss=0.03095, over 14230.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.1414, pruned_loss=0.03413, over 1727107.73 frames. ], batch size: 69, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:24:24,452 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128976.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:25:06,979 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.168e+02 2.627e+02 3.251e+02 6.252e+02, threshold=5.254e+02, percent-clipped=2.0 2022-12-08 12:25:07,063 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129024.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:25:25,477 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8707, 1.9437, 2.0874, 1.4970, 1.5116, 1.8515, 1.3020, 1.9497], device='cuda:1'), covar=tensor([0.1322, 0.1870, 0.0852, 0.2297, 0.2572, 0.1047, 0.2534, 0.0995], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0102, 0.0096, 0.0100, 0.0116, 0.0091, 0.0117, 0.0094], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:25:34,886 INFO [train.py:873] (1/4) Epoch 18, batch 500, loss[loss=0.1348, simple_loss=0.1629, pruned_loss=0.05333, over 9428.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1424, pruned_loss=0.0343, over 1832536.80 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:25:43,002 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2624, 4.3147, 4.6028, 3.8930, 4.4376, 4.6011, 1.6285, 4.1746], device='cuda:1'), covar=tensor([0.0338, 0.0332, 0.0341, 0.0493, 0.0301, 0.0251, 0.3123, 0.0311], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0177, 0.0145, 0.0148, 0.0207, 0.0143, 0.0158, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:25:47,363 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129069.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:25:57,860 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2937, 2.0178, 2.2925, 2.3935, 2.2652, 1.8784, 2.3160, 2.1839], device='cuda:1'), covar=tensor([0.0460, 0.0966, 0.0517, 0.0491, 0.0640, 0.1306, 0.0580, 0.0635], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0260, 0.0377, 0.0333, 0.0271, 0.0307, 0.0311, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:26:29,816 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:26:35,673 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.085e+02 2.575e+02 3.232e+02 8.773e+02, threshold=5.149e+02, percent-clipped=4.0 2022-12-08 12:26:47,849 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129137.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:26:59,657 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 12:27:03,407 INFO [train.py:873] (1/4) Epoch 18, batch 600, loss[loss=0.1123, simple_loss=0.1215, pruned_loss=0.05157, over 1335.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.1427, pruned_loss=0.03452, over 1927606.47 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:28:04,833 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.209e+01 2.130e+02 2.657e+02 3.284e+02 6.073e+02, threshold=5.313e+02, percent-clipped=6.0 2022-12-08 12:28:24,041 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129246.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:28:32,542 INFO [train.py:873] (1/4) Epoch 18, batch 700, loss[loss=0.1087, simple_loss=0.1308, pruned_loss=0.04331, over 3871.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1428, pruned_loss=0.03482, over 2010858.85 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 4.0 2022-12-08 12:29:33,261 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.087e+02 1.934e+02 2.414e+02 2.791e+02 7.591e+02, threshold=4.829e+02, percent-clipped=2.0 2022-12-08 12:29:59,450 INFO [train.py:873] (1/4) Epoch 18, batch 800, loss[loss=0.1241, simple_loss=0.1585, pruned_loss=0.04481, over 13988.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1425, pruned_loss=0.03492, over 1994845.90 frames. ], batch size: 22, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:30:17,127 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.50 vs. limit=5.0 2022-12-08 12:30:19,514 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7871, 1.6253, 1.8815, 1.5864, 1.7110, 1.6151, 1.4814, 1.2521], device='cuda:1'), covar=tensor([0.0158, 0.0255, 0.0164, 0.0183, 0.0166, 0.0266, 0.0236, 0.0341], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0022, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:30:51,401 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129413.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:30:54,655 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1604, 2.0480, 2.1250, 2.2242, 2.0990, 2.1109, 2.2421, 1.9214], device='cuda:1'), covar=tensor([0.1077, 0.1414, 0.0833, 0.0838, 0.1068, 0.0684, 0.1013, 0.0770], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0275, 0.0200, 0.0193, 0.0184, 0.0157, 0.0284, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:31:01,565 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.286e+02 2.776e+02 3.751e+02 1.143e+03, threshold=5.551e+02, percent-clipped=10.0 2022-12-08 12:31:11,964 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:31:27,670 INFO [train.py:873] (1/4) Epoch 18, batch 900, loss[loss=0.1056, simple_loss=0.1252, pruned_loss=0.04298, over 3879.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1426, pruned_loss=0.03507, over 1990510.66 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:31:36,180 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2025, 2.1643, 4.2627, 2.8526, 4.0042, 2.1138, 3.0481, 4.0299], device='cuda:1'), covar=tensor([0.0576, 0.3858, 0.0397, 0.5526, 0.0777, 0.3048, 0.1426, 0.0553], device='cuda:1'), in_proj_covar=tensor([0.0254, 0.0199, 0.0219, 0.0269, 0.0238, 0.0204, 0.0203, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 12:31:44,521 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129474.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 12:31:53,759 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129485.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:32:08,596 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2117, 3.9307, 3.8620, 4.2573, 4.0091, 3.7691, 4.2749, 3.5846], device='cuda:1'), covar=tensor([0.0492, 0.0928, 0.0483, 0.0435, 0.0775, 0.1119, 0.0544, 0.0560], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0278, 0.0202, 0.0196, 0.0186, 0.0159, 0.0287, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:32:28,510 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 12:32:28,737 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 2.020e+02 2.482e+02 3.043e+02 9.483e+02, threshold=4.965e+02, percent-clipped=2.0 2022-12-08 12:32:48,044 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129546.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:32:55,375 INFO [train.py:873] (1/4) Epoch 18, batch 1000, loss[loss=0.1116, simple_loss=0.1473, pruned_loss=0.03796, over 11171.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.1422, pruned_loss=0.03479, over 1982781.42 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:33:12,993 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6087, 2.4395, 2.9283, 1.8589, 1.9021, 2.5205, 1.4632, 2.5480], device='cuda:1'), covar=tensor([0.0890, 0.1177, 0.0733, 0.2445, 0.2349, 0.1085, 0.3131, 0.0937], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0101, 0.0095, 0.0099, 0.0114, 0.0091, 0.0117, 0.0093], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:33:26,834 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8311, 5.3467, 5.2647, 5.8125, 5.3662, 4.8057, 5.7684, 4.7994], device='cuda:1'), covar=tensor([0.0285, 0.0836, 0.0346, 0.0392, 0.0753, 0.0347, 0.0478, 0.0467], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0278, 0.0202, 0.0196, 0.0186, 0.0160, 0.0288, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:33:29,304 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129594.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:33:45,446 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129612.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:33:46,339 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2039, 2.0818, 2.4059, 2.1494, 2.0284, 1.9607, 1.7182, 1.7892], device='cuda:1'), covar=tensor([0.0244, 0.0345, 0.0276, 0.0362, 0.0310, 0.0457, 0.0452, 0.0483], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0022, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:33:56,802 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.171e+02 2.641e+02 3.416e+02 7.976e+02, threshold=5.283e+02, percent-clipped=6.0 2022-12-08 12:34:23,014 INFO [train.py:873] (1/4) Epoch 18, batch 1100, loss[loss=0.1421, simple_loss=0.1657, pruned_loss=0.05922, over 8618.00 frames. ], tot_loss[loss=0.1058, simple_loss=0.1421, pruned_loss=0.03471, over 1989218.59 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:34:36,303 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-08 12:34:38,287 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3948, 3.4480, 3.6771, 3.2922, 3.5723, 3.3605, 1.5250, 3.3501], device='cuda:1'), covar=tensor([0.0482, 0.0427, 0.0354, 0.0532, 0.0343, 0.0615, 0.3045, 0.0309], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0177, 0.0146, 0.0149, 0.0207, 0.0143, 0.0158, 0.0194], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:34:39,242 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129673.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:35:02,035 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 12:35:06,351 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6658, 2.6973, 2.8002, 2.7529, 2.7718, 2.4791, 1.5034, 2.5273], device='cuda:1'), covar=tensor([0.0570, 0.0509, 0.0426, 0.0421, 0.0400, 0.1088, 0.2547, 0.0389], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0177, 0.0147, 0.0149, 0.0208, 0.0143, 0.0159, 0.0195], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:35:15,775 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 12:35:20,389 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8034, 3.5463, 3.3676, 3.5100, 3.7333, 3.7756, 3.7922, 3.7907], device='cuda:1'), covar=tensor([0.0872, 0.0638, 0.1907, 0.2453, 0.0742, 0.0842, 0.0879, 0.0838], device='cuda:1'), in_proj_covar=tensor([0.0390, 0.0278, 0.0451, 0.0571, 0.0351, 0.0453, 0.0393, 0.0401], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:35:23,666 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.155e+02 2.731e+02 3.358e+02 7.442e+02, threshold=5.463e+02, percent-clipped=7.0 2022-12-08 12:35:50,418 INFO [train.py:873] (1/4) Epoch 18, batch 1200, loss[loss=0.1381, simple_loss=0.1523, pruned_loss=0.06196, over 3862.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.1413, pruned_loss=0.03395, over 2016523.84 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:36:02,201 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129769.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:36:21,547 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129791.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:36:51,110 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.078e+02 2.653e+02 3.269e+02 7.697e+02, threshold=5.307e+02, percent-clipped=3.0 2022-12-08 12:37:14,893 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:37:17,147 INFO [train.py:873] (1/4) Epoch 18, batch 1300, loss[loss=0.1306, simple_loss=0.1387, pruned_loss=0.06122, over 2659.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1423, pruned_loss=0.03528, over 1956637.78 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:37:52,362 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.7147, 5.5632, 5.2318, 5.7571, 5.2534, 5.1425, 5.8456, 5.5227], device='cuda:1'), covar=tensor([0.0666, 0.0830, 0.0822, 0.0464, 0.0687, 0.0505, 0.0531, 0.0646], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0147, 0.0150, 0.0166, 0.0150, 0.0125, 0.0172, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:38:04,603 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 12:38:18,569 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.020e+02 2.586e+02 3.074e+02 6.619e+02, threshold=5.173e+02, percent-clipped=1.0 2022-12-08 12:38:44,883 INFO [train.py:873] (1/4) Epoch 18, batch 1400, loss[loss=0.1138, simple_loss=0.1358, pruned_loss=0.04592, over 3828.00 frames. ], tot_loss[loss=0.1068, simple_loss=0.1426, pruned_loss=0.0355, over 1942360.80 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:38:45,968 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8456, 2.4391, 2.5933, 1.6884, 2.3254, 2.6166, 2.8271, 2.3084], device='cuda:1'), covar=tensor([0.0679, 0.0687, 0.0871, 0.1421, 0.1067, 0.0656, 0.0623, 0.1317], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0172, 0.0141, 0.0125, 0.0144, 0.0155, 0.0136, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 12:38:55,873 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129968.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:10,561 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1415, 2.5108, 4.1669, 4.3221, 4.1171, 2.4851, 4.2707, 3.1459], device='cuda:1'), covar=tensor([0.0444, 0.1369, 0.0973, 0.0534, 0.0537, 0.2108, 0.0518, 0.1032], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0261, 0.0378, 0.0333, 0.0273, 0.0310, 0.0313, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:39:11,447 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6305, 2.7189, 4.6371, 4.6786, 4.6179, 2.7111, 4.7481, 3.5511], device='cuda:1'), covar=tensor([0.0378, 0.1208, 0.0797, 0.0361, 0.0401, 0.1952, 0.0333, 0.0920], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0261, 0.0378, 0.0333, 0.0273, 0.0310, 0.0313, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:39:28,252 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130000.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:30,089 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:51,057 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 2.021e+02 2.494e+02 2.988e+02 5.995e+02, threshold=4.988e+02, percent-clipped=3.0 2022-12-08 12:39:57,507 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 12:40:15,733 INFO [train.py:873] (1/4) Epoch 18, batch 1500, loss[loss=0.1262, simple_loss=0.131, pruned_loss=0.06067, over 1223.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1427, pruned_loss=0.03535, over 2001902.83 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 4.0 2022-12-08 12:40:21,267 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130061.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:40:23,121 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130063.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:40:28,175 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130069.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:41:09,685 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130117.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:41:17,026 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.103e+02 2.601e+02 2.993e+02 6.866e+02, threshold=5.203e+02, percent-clipped=5.0 2022-12-08 12:41:35,571 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:41:42,391 INFO [train.py:873] (1/4) Epoch 18, batch 1600, loss[loss=0.09956, simple_loss=0.1169, pruned_loss=0.04111, over 3865.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1426, pruned_loss=0.03527, over 2024712.08 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:41:44,229 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8043, 4.5448, 4.3093, 4.8180, 4.5702, 4.2373, 4.8574, 3.9774], device='cuda:1'), covar=tensor([0.0379, 0.0823, 0.0393, 0.0405, 0.0675, 0.0680, 0.0464, 0.0508], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0276, 0.0200, 0.0194, 0.0185, 0.0159, 0.0288, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:41:48,374 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0718, 1.9244, 1.6911, 1.7207, 2.0268, 1.9812, 2.0407, 1.9873], device='cuda:1'), covar=tensor([0.1373, 0.1426, 0.3077, 0.3414, 0.1524, 0.1736, 0.1966, 0.1554], device='cuda:1'), in_proj_covar=tensor([0.0389, 0.0277, 0.0452, 0.0570, 0.0354, 0.0455, 0.0393, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:42:40,561 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7236, 3.4766, 3.3519, 3.6776, 3.5003, 3.6661, 3.7530, 3.1372], device='cuda:1'), covar=tensor([0.0398, 0.0785, 0.0463, 0.0424, 0.0715, 0.0339, 0.0493, 0.0508], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0276, 0.0199, 0.0194, 0.0185, 0.0158, 0.0287, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:42:43,871 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2022-12-08 12:42:44,998 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.040e+02 2.513e+02 3.158e+02 8.589e+02, threshold=5.027e+02, percent-clipped=3.0 2022-12-08 12:42:51,100 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6592, 1.5983, 1.7307, 1.4949, 1.3515, 1.3482, 1.5793, 1.2204], device='cuda:1'), covar=tensor([0.0166, 0.0269, 0.0160, 0.0195, 0.0233, 0.0331, 0.0215, 0.0392], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0022, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:43:10,261 INFO [train.py:873] (1/4) Epoch 18, batch 1700, loss[loss=0.1267, simple_loss=0.1322, pruned_loss=0.0606, over 1248.00 frames. ], tot_loss[loss=0.106, simple_loss=0.1423, pruned_loss=0.03485, over 2011298.64 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:43:22,102 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:44:04,000 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130316.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:44:12,322 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.118e+02 2.048e+02 2.469e+02 3.168e+02 7.190e+02, threshold=4.938e+02, percent-clipped=8.0 2022-12-08 12:44:23,812 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3749, 2.3305, 2.2747, 2.1549, 2.3469, 1.9780, 1.8428, 1.8787], device='cuda:1'), covar=tensor([0.0283, 0.0542, 0.0402, 0.0379, 0.0256, 0.0359, 0.0442, 0.0745], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0022, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:44:25,787 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.48 vs. limit=5.0 2022-12-08 12:44:37,342 INFO [train.py:873] (1/4) Epoch 18, batch 1800, loss[loss=0.1457, simple_loss=0.1477, pruned_loss=0.07183, over 1276.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1425, pruned_loss=0.03517, over 1970606.44 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:44:38,378 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:44:39,737 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 12:44:40,129 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130358.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:45:03,096 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:24,490 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-08 12:45:30,862 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 12:45:31,224 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5565, 1.4757, 3.5666, 1.5674, 3.5090, 3.6816, 2.5380, 3.9424], device='cuda:1'), covar=tensor([0.0259, 0.3209, 0.0444, 0.2359, 0.0682, 0.0437, 0.0908, 0.0195], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0156, 0.0160, 0.0169, 0.0169, 0.0180, 0.0134, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:45:37,729 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3348, 2.1614, 3.1734, 2.4670, 3.0986, 3.0505, 2.9797, 2.6235], device='cuda:1'), covar=tensor([0.0973, 0.3032, 0.1275, 0.1932, 0.1009, 0.1153, 0.1290, 0.1798], device='cuda:1'), in_proj_covar=tensor([0.0355, 0.0312, 0.0392, 0.0295, 0.0366, 0.0321, 0.0359, 0.0299], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:45:39,183 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.710e+01 2.001e+02 2.479e+02 3.133e+02 6.689e+02, threshold=4.957e+02, percent-clipped=4.0 2022-12-08 12:45:43,997 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 12:45:44,488 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:49,402 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130438.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:55,449 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:57,209 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130447.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:46:04,198 INFO [train.py:873] (1/4) Epoch 18, batch 1900, loss[loss=0.1093, simple_loss=0.1464, pruned_loss=0.03614, over 14487.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1425, pruned_loss=0.03494, over 1999493.58 frames. ], batch size: 49, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:46:08,807 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4433, 2.4675, 2.5557, 2.4934, 2.4976, 2.1084, 1.5131, 2.2514], device='cuda:1'), covar=tensor([0.0638, 0.0497, 0.0469, 0.0410, 0.0431, 0.1253, 0.2309, 0.0410], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0178, 0.0147, 0.0150, 0.0210, 0.0144, 0.0159, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:46:37,684 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:46:39,402 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:46:42,963 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130499.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:47:06,079 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.089e+02 2.666e+02 3.206e+02 5.999e+02, threshold=5.333e+02, percent-clipped=5.0 2022-12-08 12:47:19,940 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 12:47:25,689 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:47:31,376 INFO [train.py:873] (1/4) Epoch 18, batch 2000, loss[loss=0.1243, simple_loss=0.1434, pruned_loss=0.05255, over 3866.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.1418, pruned_loss=0.03522, over 1883957.45 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:47:40,354 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130565.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:48:09,828 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6157, 3.9888, 3.2862, 4.9187, 4.3239, 4.7189, 4.1032, 3.4354], device='cuda:1'), covar=tensor([0.0630, 0.1172, 0.3010, 0.0428, 0.0952, 0.1328, 0.1131, 0.2658], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0285, 0.0257, 0.0288, 0.0319, 0.0299, 0.0250, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:48:18,839 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130609.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:48:19,755 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6405, 1.7187, 1.7522, 1.6594, 1.5506, 1.4749, 1.4492, 1.1635], device='cuda:1'), covar=tensor([0.0207, 0.0266, 0.0202, 0.0184, 0.0213, 0.0333, 0.0247, 0.0366], device='cuda:1'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0032], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:48:24,904 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 12:48:34,556 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:48:35,244 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.007e+02 2.517e+02 3.001e+02 4.503e+02, threshold=5.034e+02, percent-clipped=0.0 2022-12-08 12:48:59,205 INFO [train.py:873] (1/4) Epoch 18, batch 2100, loss[loss=0.1203, simple_loss=0.1422, pruned_loss=0.04919, over 4930.00 frames. ], tot_loss[loss=0.1058, simple_loss=0.1416, pruned_loss=0.03505, over 1940769.99 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:49:00,415 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130656.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:49:02,034 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130658.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 12:49:41,919 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:49:43,626 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130706.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:49:58,857 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-08 12:50:00,944 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0865, 4.9491, 4.6533, 5.0863, 4.6867, 4.3836, 5.1621, 4.8824], device='cuda:1'), covar=tensor([0.0550, 0.0631, 0.0733, 0.0460, 0.0547, 0.0585, 0.0509, 0.0596], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0147, 0.0150, 0.0165, 0.0151, 0.0126, 0.0172, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:50:00,983 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0771, 4.1533, 4.2465, 3.9974, 4.1849, 4.3768, 1.6716, 3.8761], device='cuda:1'), covar=tensor([0.0414, 0.0364, 0.0530, 0.0504, 0.0409, 0.0302, 0.3686, 0.0394], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0176, 0.0146, 0.0149, 0.0208, 0.0143, 0.0158, 0.0195], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 12:50:01,668 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.249e+02 2.636e+02 3.244e+02 1.045e+03, threshold=5.271e+02, percent-clipped=2.0 2022-12-08 12:50:13,056 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:50:26,107 INFO [train.py:873] (1/4) Epoch 18, batch 2200, loss[loss=0.1639, simple_loss=0.1778, pruned_loss=0.07497, over 8594.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1417, pruned_loss=0.03545, over 1894168.32 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:50:55,182 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:50:57,839 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5897, 2.3066, 3.4860, 2.7094, 3.4816, 3.3912, 3.2882, 2.9387], device='cuda:1'), covar=tensor([0.1003, 0.2904, 0.1040, 0.1754, 0.0803, 0.1054, 0.1259, 0.1586], device='cuda:1'), in_proj_covar=tensor([0.0354, 0.0309, 0.0390, 0.0293, 0.0364, 0.0320, 0.0358, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 12:51:00,296 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130794.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:51:29,862 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.113e+02 2.522e+02 3.160e+02 5.458e+02, threshold=5.044e+02, percent-clipped=1.0 2022-12-08 12:51:31,803 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9116, 4.1630, 3.8501, 3.8321, 2.9610, 4.0330, 3.9523, 2.4065], device='cuda:1'), covar=tensor([0.1128, 0.0546, 0.1067, 0.0895, 0.0797, 0.0660, 0.0729, 0.1675], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0091, 0.0071, 0.0077, 0.0101, 0.0091, 0.0102, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 12:51:42,756 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-08 12:51:53,881 INFO [train.py:873] (1/4) Epoch 18, batch 2300, loss[loss=0.1207, simple_loss=0.1443, pruned_loss=0.04852, over 5961.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1414, pruned_loss=0.03501, over 1955841.27 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:52:37,094 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130904.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:52:51,892 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130921.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:52:56,900 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.034e+02 2.487e+02 3.081e+02 6.323e+02, threshold=4.975e+02, percent-clipped=2.0 2022-12-08 12:53:10,227 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.76 vs. limit=5.0 2022-12-08 12:53:22,074 INFO [train.py:873] (1/4) Epoch 18, batch 2400, loss[loss=0.1133, simple_loss=0.1506, pruned_loss=0.03804, over 14291.00 frames. ], tot_loss[loss=0.1058, simple_loss=0.1413, pruned_loss=0.03512, over 1888855.95 frames. ], batch size: 60, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:54:25,831 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.032e+02 2.544e+02 3.351e+02 8.852e+02, threshold=5.088e+02, percent-clipped=2.0 2022-12-08 12:54:35,198 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.47 vs. limit=5.0 2022-12-08 12:54:37,311 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:54:50,162 INFO [train.py:873] (1/4) Epoch 18, batch 2500, loss[loss=0.1147, simple_loss=0.1522, pruned_loss=0.03859, over 14241.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1411, pruned_loss=0.03514, over 1853703.32 frames. ], batch size: 37, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:55:19,755 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:19,852 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:25,071 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131094.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:34,632 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9579, 3.7482, 3.6186, 3.9947, 3.7917, 3.6181, 4.0531, 3.3447], device='cuda:1'), covar=tensor([0.0543, 0.0976, 0.0521, 0.0473, 0.0843, 0.1325, 0.0565, 0.0611], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0279, 0.0201, 0.0196, 0.0184, 0.0159, 0.0290, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 12:55:43,359 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.50 vs. limit=5.0 2022-12-08 12:55:43,979 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5117, 2.4019, 3.0100, 1.9029, 2.1237, 2.7249, 1.6007, 2.6670], device='cuda:1'), covar=tensor([0.0928, 0.1308, 0.0774, 0.1978, 0.1971, 0.0858, 0.3070, 0.0927], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0104, 0.0096, 0.0101, 0.0116, 0.0093, 0.0119, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 12:55:54,048 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.009e+02 2.192e+02 2.699e+02 3.352e+02 8.933e+02, threshold=5.398e+02, percent-clipped=5.0 2022-12-08 12:56:01,760 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:56:07,942 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:56:19,377 INFO [train.py:873] (1/4) Epoch 18, batch 2600, loss[loss=0.1101, simple_loss=0.1471, pruned_loss=0.03652, over 13927.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1414, pruned_loss=0.03472, over 1938436.02 frames. ], batch size: 23, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:56:35,579 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131174.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:56:37,718 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 12:57:02,643 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131204.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:57:17,441 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:57:20,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-08 12:57:20,812 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0565, 3.9324, 3.7143, 4.0904, 3.6556, 3.4491, 4.0914, 3.9204], device='cuda:1'), covar=tensor([0.0654, 0.0818, 0.0910, 0.0564, 0.0951, 0.0722, 0.0647, 0.0792], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0166, 0.0152, 0.0127, 0.0175, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 12:57:22,469 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 2.143e+02 2.636e+02 3.180e+02 5.935e+02, threshold=5.272e+02, percent-clipped=1.0 2022-12-08 12:57:29,937 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131235.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:57:35,930 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 12:57:44,012 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131252.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:57:46,565 INFO [train.py:873] (1/4) Epoch 18, batch 2700, loss[loss=0.1015, simple_loss=0.144, pruned_loss=0.02952, over 14246.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1414, pruned_loss=0.03418, over 2005808.96 frames. ], batch size: 60, lr: 4.38e-03, grad_scale: 4.0 2022-12-08 12:57:57,920 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.30 vs. limit=5.0 2022-12-08 12:57:59,023 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:58:02,021 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8074, 1.3874, 2.7367, 2.4620, 2.6362, 2.6929, 2.0074, 2.7125], device='cuda:1'), covar=tensor([0.1342, 0.1595, 0.0228, 0.0457, 0.0474, 0.0254, 0.0616, 0.0281], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0132, 0.0169, 0.0149, 0.0145, 0.0127, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 12:58:50,252 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.346e+01 2.091e+02 2.717e+02 3.519e+02 6.308e+02, threshold=5.434e+02, percent-clipped=4.0 2022-12-08 12:59:00,846 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2022-12-08 12:59:06,327 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1123, 1.0951, 0.9865, 1.0717, 1.2147, 0.6647, 0.9480, 1.1435], device='cuda:1'), covar=tensor([0.0731, 0.0764, 0.0691, 0.0551, 0.0516, 0.0667, 0.0997, 0.0759], device='cuda:1'), in_proj_covar=tensor([0.0037, 0.0035, 0.0040, 0.0033, 0.0034, 0.0048, 0.0037, 0.0039], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 12:59:14,298 INFO [train.py:873] (1/4) Epoch 18, batch 2800, loss[loss=0.133, simple_loss=0.1392, pruned_loss=0.06342, over 1168.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1416, pruned_loss=0.03414, over 1970938.78 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:59:47,106 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1275, 1.9838, 2.0972, 2.1829, 2.0938, 2.0128, 2.2029, 1.8901], device='cuda:1'), covar=tensor([0.1215, 0.1444, 0.0816, 0.0855, 0.0968, 0.0788, 0.0864, 0.0803], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0279, 0.0202, 0.0197, 0.0184, 0.0160, 0.0291, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 13:00:02,553 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131409.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:00:08,542 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131416.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:00:18,983 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.304e+02 1.962e+02 2.375e+02 2.873e+02 4.439e+02, threshold=4.749e+02, percent-clipped=0.0 2022-12-08 13:00:27,085 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1728, 1.3438, 3.2271, 1.4979, 3.0757, 3.2461, 2.3000, 3.4950], device='cuda:1'), covar=tensor([0.0288, 0.3304, 0.0408, 0.2291, 0.1002, 0.0451, 0.1086, 0.0244], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0154, 0.0158, 0.0166, 0.0167, 0.0177, 0.0132, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:00:30,092 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 13:00:33,986 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1936, 2.8181, 2.8884, 1.8587, 2.6802, 2.8512, 3.1903, 2.5906], device='cuda:1'), covar=tensor([0.0622, 0.0741, 0.0848, 0.1341, 0.0885, 0.0572, 0.0587, 0.1049], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0170, 0.0141, 0.0126, 0.0143, 0.0155, 0.0136, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 13:00:42,310 INFO [train.py:873] (1/4) Epoch 18, batch 2900, loss[loss=0.0848, simple_loss=0.1318, pruned_loss=0.01892, over 11281.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1416, pruned_loss=0.03406, over 2025241.19 frames. ], batch size: 14, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 13:00:55,819 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131470.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:02,120 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131477.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:15,227 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131492.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:46,458 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.257e+02 2.729e+02 3.510e+02 6.783e+02, threshold=5.458e+02, percent-clipped=7.0 2022-12-08 13:01:48,382 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131530.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 13:02:08,719 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131553.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:02:10,258 INFO [train.py:873] (1/4) Epoch 18, batch 3000, loss[loss=0.1008, simple_loss=0.1088, pruned_loss=0.04643, over 2592.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1419, pruned_loss=0.03444, over 2023416.64 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 13:02:10,258 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 13:02:18,733 INFO [train.py:905] (1/4) Epoch 18, validation: loss=0.1388, simple_loss=0.176, pruned_loss=0.05082, over 857387.00 frames. 2022-12-08 13:02:18,734 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 13:02:59,575 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-08 13:03:22,346 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 2.067e+02 2.535e+02 3.249e+02 7.779e+02, threshold=5.070e+02, percent-clipped=4.0 2022-12-08 13:03:44,986 INFO [train.py:873] (1/4) Epoch 18, batch 3100, loss[loss=0.1037, simple_loss=0.1451, pruned_loss=0.03112, over 14661.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1418, pruned_loss=0.03435, over 2009417.35 frames. ], batch size: 33, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:04:15,087 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8999, 1.2090, 2.0038, 1.2436, 1.9406, 2.0534, 1.5636, 2.1188], device='cuda:1'), covar=tensor([0.0314, 0.2471, 0.0576, 0.2108, 0.0667, 0.0648, 0.1362, 0.0468], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0156, 0.0160, 0.0167, 0.0169, 0.0178, 0.0134, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:04:38,598 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2022-12-08 13:04:47,492 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-08 13:04:49,757 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 2.137e+02 2.592e+02 3.013e+02 4.856e+02, threshold=5.183e+02, percent-clipped=0.0 2022-12-08 13:05:12,289 INFO [train.py:873] (1/4) Epoch 18, batch 3200, loss[loss=0.1053, simple_loss=0.137, pruned_loss=0.03675, over 6969.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1418, pruned_loss=0.03458, over 2002831.46 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:05:20,637 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:05:27,341 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:05:34,201 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8177, 1.6768, 3.9200, 3.5691, 3.6519, 3.9621, 3.3201, 3.9243], device='cuda:1'), covar=tensor([0.1745, 0.1742, 0.0142, 0.0320, 0.0277, 0.0155, 0.0272, 0.0156], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0158, 0.0131, 0.0168, 0.0147, 0.0143, 0.0126, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 13:05:45,169 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.40 vs. limit=5.0 2022-12-08 13:05:52,825 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6309, 2.3522, 2.4617, 2.4960, 2.2488, 1.5262, 2.2324, 2.7544], device='cuda:1'), covar=tensor([0.0859, 0.0621, 0.0585, 0.0799, 0.0973, 0.0800, 0.1524, 0.0596], device='cuda:1'), in_proj_covar=tensor([0.0036, 0.0034, 0.0039, 0.0033, 0.0034, 0.0048, 0.0036, 0.0038], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 13:06:17,105 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.113e+02 2.491e+02 2.980e+02 5.007e+02, threshold=4.982e+02, percent-clipped=0.0 2022-12-08 13:06:18,128 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:06:19,733 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8165, 1.5435, 1.8462, 1.9769, 1.4216, 1.7299, 1.7062, 1.8368], device='cuda:1'), covar=tensor([0.0273, 0.0392, 0.0221, 0.0225, 0.0480, 0.0490, 0.0341, 0.0240], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0260, 0.0376, 0.0331, 0.0271, 0.0307, 0.0312, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:06:34,164 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131848.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:06:40,047 INFO [train.py:873] (1/4) Epoch 18, batch 3300, loss[loss=0.09104, simple_loss=0.1308, pruned_loss=0.02565, over 13977.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1409, pruned_loss=0.03369, over 2014377.23 frames. ], batch size: 19, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:06:41,832 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131857.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:00,063 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131878.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:34,798 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:44,548 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.221e+02 2.033e+02 2.408e+02 3.060e+02 5.712e+02, threshold=4.815e+02, percent-clipped=3.0 2022-12-08 13:08:06,596 INFO [train.py:873] (1/4) Epoch 18, batch 3400, loss[loss=0.09855, simple_loss=0.1376, pruned_loss=0.02974, over 11214.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.1408, pruned_loss=0.03384, over 1933309.09 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:08:46,518 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0174, 2.9149, 2.2623, 3.0999, 2.9525, 2.9413, 2.7116, 2.3245], device='cuda:1'), covar=tensor([0.1047, 0.1275, 0.3150, 0.0855, 0.1129, 0.1452, 0.1452, 0.2786], device='cuda:1'), in_proj_covar=tensor([0.0286, 0.0293, 0.0260, 0.0294, 0.0326, 0.0305, 0.0255, 0.0245], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:09:01,183 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4336, 4.1128, 3.9141, 4.0671, 4.2679, 4.3079, 4.3503, 4.3781], device='cuda:1'), covar=tensor([0.0679, 0.0647, 0.1995, 0.2494, 0.0721, 0.0814, 0.0824, 0.0762], device='cuda:1'), in_proj_covar=tensor([0.0393, 0.0278, 0.0449, 0.0570, 0.0350, 0.0452, 0.0393, 0.0400], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:09:10,665 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.058e+02 2.441e+02 2.998e+02 5.403e+02, threshold=4.883e+02, percent-clipped=4.0 2022-12-08 13:09:12,607 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132031.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:30,552 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-08 13:09:32,536 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132054.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:33,283 INFO [train.py:873] (1/4) Epoch 18, batch 3500, loss[loss=0.1301, simple_loss=0.1597, pruned_loss=0.05027, over 8609.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1405, pruned_loss=0.03393, over 1928125.53 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:09:42,168 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132065.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:47,799 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:52,117 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9255, 1.9955, 2.1774, 1.4702, 1.6298, 2.0286, 1.3613, 2.0477], device='cuda:1'), covar=tensor([0.1027, 0.1758, 0.0852, 0.2482, 0.2336, 0.1031, 0.2854, 0.0996], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0103, 0.0096, 0.0101, 0.0116, 0.0092, 0.0118, 0.0095], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 13:10:05,396 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:23,462 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132113.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:25,281 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132115.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 13:10:26,350 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 13:10:29,237 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132120.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:33,693 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 13:10:37,536 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.039e+02 2.680e+02 3.337e+02 6.531e+02, threshold=5.359e+02, percent-clipped=3.0 2022-12-08 13:10:39,274 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8514, 0.7750, 0.8680, 0.7986, 0.7427, 0.5316, 0.6406, 0.6895], device='cuda:1'), covar=tensor([0.0186, 0.0183, 0.0169, 0.0184, 0.0205, 0.0344, 0.0213, 0.0285], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0023, 0.0021, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 13:10:54,090 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132148.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:59,847 INFO [train.py:873] (1/4) Epoch 18, batch 3600, loss[loss=0.1151, simple_loss=0.1372, pruned_loss=0.04654, over 3828.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.1411, pruned_loss=0.03508, over 1803859.11 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:11:35,849 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132196.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:11:51,501 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:12:05,413 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.046e+02 2.523e+02 3.076e+02 9.449e+02, threshold=5.046e+02, percent-clipped=2.0 2022-12-08 13:12:06,914 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 13:12:10,513 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3274, 4.1582, 4.0075, 4.3019, 3.9662, 3.7440, 4.3715, 4.1133], device='cuda:1'), covar=tensor([0.0563, 0.0834, 0.0788, 0.0586, 0.0822, 0.0672, 0.0554, 0.0766], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0145, 0.0148, 0.0164, 0.0148, 0.0124, 0.0171, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:12:11,814 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2022-12-08 13:12:28,035 INFO [train.py:873] (1/4) Epoch 18, batch 3700, loss[loss=0.1493, simple_loss=0.138, pruned_loss=0.08035, over 1227.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1413, pruned_loss=0.0347, over 1876084.41 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:12:45,200 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 13:12:59,545 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8656, 4.7534, 4.4473, 4.8173, 4.4516, 4.3354, 4.9332, 4.5915], device='cuda:1'), covar=tensor([0.0595, 0.0822, 0.0790, 0.0594, 0.0708, 0.0575, 0.0584, 0.0803], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0165, 0.0149, 0.0125, 0.0172, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:13:05,645 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9253, 1.7733, 1.8576, 1.9248, 2.0106, 1.1805, 1.6533, 1.8218], device='cuda:1'), covar=tensor([0.0714, 0.0847, 0.0794, 0.0720, 0.0367, 0.0864, 0.0829, 0.0964], device='cuda:1'), in_proj_covar=tensor([0.0037, 0.0035, 0.0040, 0.0033, 0.0035, 0.0049, 0.0036, 0.0039], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:1') 2022-12-08 13:13:32,947 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.108e+02 2.462e+02 3.226e+02 6.885e+02, threshold=4.924e+02, percent-clipped=3.0 2022-12-08 13:13:54,142 INFO [train.py:873] (1/4) Epoch 18, batch 3800, loss[loss=0.1002, simple_loss=0.1403, pruned_loss=0.02999, over 6920.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1406, pruned_loss=0.03439, over 1896011.22 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 4.0 2022-12-08 13:14:10,625 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7205, 1.2163, 3.0583, 1.5683, 2.9277, 2.9674, 2.2611, 2.9314], device='cuda:1'), covar=tensor([0.0545, 0.4356, 0.0628, 0.2794, 0.0768, 0.0711, 0.1036, 0.0678], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0160, 0.0167, 0.0168, 0.0179, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:14:14,325 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 13:14:22,238 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132387.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:14:42,689 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132410.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 13:14:59,235 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.016e+02 2.539e+02 3.051e+02 5.858e+02, threshold=5.078e+02, percent-clipped=3.0 2022-12-08 13:15:21,337 INFO [train.py:873] (1/4) Epoch 18, batch 3900, loss[loss=0.08957, simple_loss=0.132, pruned_loss=0.0236, over 14050.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1409, pruned_loss=0.03452, over 1911312.47 frames. ], batch size: 22, lr: 4.36e-03, grad_scale: 4.0 2022-12-08 13:16:11,305 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8411, 1.5457, 2.0803, 1.6267, 1.8721, 1.4359, 1.6697, 1.9621], device='cuda:1'), covar=tensor([0.2825, 0.2639, 0.0468, 0.1987, 0.1634, 0.1381, 0.1138, 0.1037], device='cuda:1'), in_proj_covar=tensor([0.0254, 0.0198, 0.0220, 0.0270, 0.0238, 0.0201, 0.0200, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 13:16:12,121 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:16:21,270 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1612, 3.9042, 3.8597, 4.1158, 3.7649, 3.5948, 4.1803, 3.9833], device='cuda:1'), covar=tensor([0.0634, 0.0920, 0.0882, 0.0701, 0.0917, 0.0673, 0.0636, 0.0775], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0145, 0.0148, 0.0164, 0.0149, 0.0124, 0.0171, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:16:26,818 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.073e+02 2.605e+02 3.188e+02 7.778e+02, threshold=5.211e+02, percent-clipped=4.0 2022-12-08 13:16:48,238 INFO [train.py:873] (1/4) Epoch 18, batch 4000, loss[loss=0.09949, simple_loss=0.1455, pruned_loss=0.02673, over 13780.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.1411, pruned_loss=0.0341, over 1932947.97 frames. ], batch size: 23, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:16:53,527 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132561.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:17:54,706 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.137e+02 2.506e+02 3.319e+02 7.426e+02, threshold=5.012e+02, percent-clipped=3.0 2022-12-08 13:18:16,274 INFO [train.py:873] (1/4) Epoch 18, batch 4100, loss[loss=0.1421, simple_loss=0.1444, pruned_loss=0.06994, over 2611.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1408, pruned_loss=0.0338, over 1958452.21 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:18:20,658 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132660.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:18:44,553 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132687.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:02,979 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 13:19:05,222 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132710.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:19:14,321 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0182, 3.9086, 3.5130, 3.6136, 4.0533, 4.0800, 4.1121, 4.0417], device='cuda:1'), covar=tensor([0.1712, 0.0779, 0.2746, 0.3794, 0.0980, 0.1346, 0.1239, 0.1505], device='cuda:1'), in_proj_covar=tensor([0.0393, 0.0275, 0.0448, 0.0569, 0.0349, 0.0450, 0.0389, 0.0398], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:19:14,431 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132721.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:22,530 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.213e+02 2.659e+02 3.609e+02 5.635e+02, threshold=5.318e+02, percent-clipped=5.0 2022-12-08 13:19:27,143 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132735.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:36,788 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132746.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:44,647 INFO [train.py:873] (1/4) Epoch 18, batch 4200, loss[loss=0.1725, simple_loss=0.1518, pruned_loss=0.09664, over 1283.00 frames. ], tot_loss[loss=0.105, simple_loss=0.1416, pruned_loss=0.03422, over 1948054.04 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:19:47,325 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132758.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:20:30,871 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132807.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:20:50,970 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.134e+02 2.571e+02 3.230e+02 5.703e+02, threshold=5.142e+02, percent-clipped=2.0 2022-12-08 13:21:04,769 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 13:21:13,200 INFO [train.py:873] (1/4) Epoch 18, batch 4300, loss[loss=0.1297, simple_loss=0.1603, pruned_loss=0.04952, over 12751.00 frames. ], tot_loss[loss=0.1052, simple_loss=0.1416, pruned_loss=0.03439, over 1984234.71 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 4.0 2022-12-08 13:21:54,061 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1375, 3.5422, 3.3334, 3.4206, 2.5659, 3.5369, 3.3732, 1.7590], device='cuda:1'), covar=tensor([0.1304, 0.0787, 0.0781, 0.1017, 0.0952, 0.0479, 0.0805, 0.2160], device='cuda:1'), in_proj_covar=tensor([0.0141, 0.0092, 0.0072, 0.0078, 0.0101, 0.0093, 0.0103, 0.0100], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 13:22:20,489 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 2.027e+02 2.464e+02 3.102e+02 5.439e+02, threshold=4.929e+02, percent-clipped=1.0 2022-12-08 13:22:41,307 INFO [train.py:873] (1/4) Epoch 18, batch 4400, loss[loss=0.1006, simple_loss=0.1286, pruned_loss=0.03632, over 4952.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1415, pruned_loss=0.03415, over 2003491.69 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:23:35,944 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133016.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:23:48,921 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.037e+02 1.881e+02 2.190e+02 2.815e+02 5.316e+02, threshold=4.380e+02, percent-clipped=1.0 2022-12-08 13:24:10,912 INFO [train.py:873] (1/4) Epoch 18, batch 4500, loss[loss=0.114, simple_loss=0.1491, pruned_loss=0.03948, over 14269.00 frames. ], tot_loss[loss=0.104, simple_loss=0.141, pruned_loss=0.03352, over 1989397.26 frames. ], batch size: 46, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:24:52,348 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133102.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:25:17,475 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.087e+02 2.404e+02 3.093e+02 4.922e+02, threshold=4.807e+02, percent-clipped=5.0 2022-12-08 13:25:38,354 INFO [train.py:873] (1/4) Epoch 18, batch 4600, loss[loss=0.1231, simple_loss=0.1473, pruned_loss=0.04951, over 9480.00 frames. ], tot_loss[loss=0.1052, simple_loss=0.1416, pruned_loss=0.03434, over 1948983.92 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:26:21,070 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5868, 4.0018, 3.2859, 4.9798, 4.4018, 4.6477, 3.9361, 3.5748], device='cuda:1'), covar=tensor([0.0618, 0.1121, 0.2750, 0.0439, 0.0773, 0.1133, 0.1152, 0.2398], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0288, 0.0259, 0.0290, 0.0320, 0.0302, 0.0253, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:26:44,620 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 2.089e+02 2.539e+02 2.980e+02 5.650e+02, threshold=5.079e+02, percent-clipped=2.0 2022-12-08 13:27:05,957 INFO [train.py:873] (1/4) Epoch 18, batch 4700, loss[loss=0.09762, simple_loss=0.1376, pruned_loss=0.02882, over 11177.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1413, pruned_loss=0.03405, over 1972334.37 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:27:58,996 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133316.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:28:12,073 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.140e+02 2.537e+02 3.389e+02 1.353e+03, threshold=5.074e+02, percent-clipped=8.0 2022-12-08 13:28:18,045 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 13:28:32,822 INFO [train.py:873] (1/4) Epoch 18, batch 4800, loss[loss=0.09892, simple_loss=0.1407, pruned_loss=0.02858, over 14643.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.1411, pruned_loss=0.03389, over 1948087.78 frames. ], batch size: 23, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:28:40,333 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=133364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:28:49,179 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133374.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 13:28:51,167 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:00,274 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 13:29:10,527 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.77 vs. limit=5.0 2022-12-08 13:29:10,579 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-08 13:29:13,882 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133402.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:17,119 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3929, 4.1308, 3.8777, 4.0240, 4.2281, 4.3097, 4.3667, 4.3202], device='cuda:1'), covar=tensor([0.0863, 0.0590, 0.2286, 0.2723, 0.0772, 0.0834, 0.0936, 0.0914], device='cuda:1'), in_proj_covar=tensor([0.0388, 0.0276, 0.0449, 0.0570, 0.0349, 0.0451, 0.0388, 0.0398], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:29:38,999 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 1.939e+02 2.430e+02 2.952e+02 7.648e+02, threshold=4.861e+02, percent-clipped=3.0 2022-12-08 13:29:42,908 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133435.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 13:29:44,591 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:55,370 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=133450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:59,644 INFO [train.py:873] (1/4) Epoch 18, batch 4900, loss[loss=0.09266, simple_loss=0.134, pruned_loss=0.02567, over 14256.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.1409, pruned_loss=0.03399, over 1939655.41 frames. ], batch size: 69, lr: 4.34e-03, grad_scale: 8.0 2022-12-08 13:30:05,064 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2022-12-08 13:30:16,291 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:30:49,094 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6381, 1.6335, 1.7012, 1.5638, 1.5029, 1.4531, 1.2902, 1.2096], device='cuda:1'), covar=tensor([0.0207, 0.0297, 0.0167, 0.0188, 0.0191, 0.0313, 0.0228, 0.0313], device='cuda:1'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 13:31:05,364 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.069e+02 2.523e+02 3.509e+02 9.363e+02, threshold=5.045e+02, percent-clipped=9.0 2022-12-08 13:31:09,289 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133535.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:31:26,063 INFO [train.py:873] (1/4) Epoch 18, batch 5000, loss[loss=0.1156, simple_loss=0.1574, pruned_loss=0.03695, over 14209.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.1413, pruned_loss=0.03368, over 1987182.47 frames. ], batch size: 94, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:31:37,766 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 13:32:32,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.004e+02 2.618e+02 3.128e+02 4.580e+02, threshold=5.235e+02, percent-clipped=0.0 2022-12-08 13:32:39,820 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 13:32:50,853 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133652.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:32:53,176 INFO [train.py:873] (1/4) Epoch 18, batch 5100, loss[loss=0.1167, simple_loss=0.156, pruned_loss=0.03873, over 14482.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1415, pruned_loss=0.03454, over 1953750.23 frames. ], batch size: 51, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:33:33,488 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9139, 2.7269, 2.7616, 2.9134, 2.8420, 2.8516, 2.9843, 2.4967], device='cuda:1'), covar=tensor([0.0620, 0.1101, 0.0623, 0.0598, 0.0735, 0.0534, 0.0710, 0.0642], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0275, 0.0200, 0.0194, 0.0183, 0.0159, 0.0290, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 13:33:44,295 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133713.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:33:58,831 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133730.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 13:34:00,754 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.056e+02 2.175e+02 2.651e+02 3.112e+02 1.000e+03, threshold=5.302e+02, percent-clipped=1.0 2022-12-08 13:34:00,867 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:34:21,229 INFO [train.py:873] (1/4) Epoch 18, batch 5200, loss[loss=0.08167, simple_loss=0.1227, pruned_loss=0.02031, over 6958.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1407, pruned_loss=0.03356, over 1983724.77 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 8.0 2022-12-08 13:34:52,833 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7147, 2.8778, 4.4513, 3.3313, 4.3376, 4.2622, 4.2551, 3.8706], device='cuda:1'), covar=tensor([0.0685, 0.3198, 0.0920, 0.1830, 0.0878, 0.1000, 0.1505, 0.1446], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0306, 0.0387, 0.0296, 0.0362, 0.0316, 0.0355, 0.0294], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:35:13,550 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2022-12-08 13:35:27,078 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:35:29,270 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 1.917e+02 2.403e+02 3.012e+02 6.543e+02, threshold=4.806e+02, percent-clipped=3.0 2022-12-08 13:35:48,359 INFO [train.py:873] (1/4) Epoch 18, batch 5300, loss[loss=0.1186, simple_loss=0.1432, pruned_loss=0.04694, over 6003.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1412, pruned_loss=0.03404, over 2004028.63 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:35:54,299 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7774, 1.4918, 2.0144, 1.6197, 1.8330, 1.3426, 1.6204, 1.9312], device='cuda:1'), covar=tensor([0.3257, 0.3493, 0.0696, 0.2068, 0.2000, 0.1631, 0.1452, 0.1048], device='cuda:1'), in_proj_covar=tensor([0.0255, 0.0200, 0.0223, 0.0269, 0.0240, 0.0203, 0.0203, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 13:36:06,378 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0913, 1.9493, 4.6609, 4.3026, 4.1985, 4.7378, 4.3399, 4.7499], device='cuda:1'), covar=tensor([0.1560, 0.1569, 0.0105, 0.0211, 0.0244, 0.0145, 0.0155, 0.0115], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0129, 0.0167, 0.0146, 0.0143, 0.0125, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 13:36:55,585 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.086e+02 2.076e+02 2.589e+02 3.060e+02 6.026e+02, threshold=5.178e+02, percent-clipped=1.0 2022-12-08 13:37:14,384 INFO [train.py:873] (1/4) Epoch 18, batch 5400, loss[loss=0.09191, simple_loss=0.1242, pruned_loss=0.02979, over 3876.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.1413, pruned_loss=0.03418, over 1995842.84 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:37:50,991 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133997.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:00,666 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:15,979 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0344, 2.1333, 1.9392, 2.1410, 1.7880, 2.0171, 2.1212, 2.0768], device='cuda:1'), covar=tensor([0.1037, 0.1124, 0.1330, 0.0957, 0.1542, 0.1028, 0.1136, 0.1073], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0147, 0.0151, 0.0166, 0.0152, 0.0126, 0.0172, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:38:20,343 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134030.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 13:38:22,056 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:22,721 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 1.858e+02 2.390e+02 3.308e+02 6.761e+02, threshold=4.781e+02, percent-clipped=5.0 2022-12-08 13:38:40,553 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2690, 2.6009, 4.1399, 3.0424, 4.0972, 3.9610, 3.9297, 3.5177], device='cuda:1'), covar=tensor([0.0814, 0.3196, 0.0934, 0.1799, 0.0847, 0.1013, 0.1629, 0.1603], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0307, 0.0387, 0.0297, 0.0361, 0.0317, 0.0356, 0.0294], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:38:42,049 INFO [train.py:873] (1/4) Epoch 18, batch 5500, loss[loss=0.1553, simple_loss=0.1458, pruned_loss=0.0824, over 1324.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1407, pruned_loss=0.03382, over 1984150.52 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 4.0 2022-12-08 13:38:44,816 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:02,240 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134078.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:39:03,893 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:43,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1297, 2.0491, 4.0931, 2.9448, 4.0090, 1.9831, 3.0584, 3.9935], device='cuda:1'), covar=tensor([0.0692, 0.4161, 0.0501, 0.5282, 0.0715, 0.3158, 0.1529, 0.0520], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0200, 0.0223, 0.0271, 0.0239, 0.0202, 0.0203, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 13:39:47,390 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:50,149 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.212e+02 2.722e+02 3.305e+02 6.043e+02, threshold=5.445e+02, percent-clipped=3.0 2022-12-08 13:39:53,848 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3803, 1.5950, 4.2047, 2.0048, 4.1580, 4.3949, 3.6708, 4.7325], device='cuda:1'), covar=tensor([0.0210, 0.2962, 0.0416, 0.2032, 0.0360, 0.0312, 0.0427, 0.0159], device='cuda:1'), in_proj_covar=tensor([0.0171, 0.0153, 0.0159, 0.0165, 0.0167, 0.0177, 0.0131, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:40:01,200 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6395, 3.2940, 2.6221, 3.7786, 3.6502, 3.6440, 3.1995, 2.7016], device='cuda:1'), covar=tensor([0.0760, 0.1366, 0.3044, 0.0656, 0.0882, 0.1106, 0.1307, 0.2862], device='cuda:1'), in_proj_covar=tensor([0.0285, 0.0291, 0.0260, 0.0294, 0.0324, 0.0305, 0.0256, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:40:09,463 INFO [train.py:873] (1/4) Epoch 18, batch 5600, loss[loss=0.1223, simple_loss=0.1541, pruned_loss=0.04524, over 10338.00 frames. ], tot_loss[loss=0.105, simple_loss=0.1411, pruned_loss=0.0345, over 1911118.24 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:40:18,203 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9758, 4.8138, 4.5457, 4.9781, 4.5934, 4.4554, 5.0102, 4.7863], device='cuda:1'), covar=tensor([0.0544, 0.0769, 0.0824, 0.0543, 0.0690, 0.0509, 0.0531, 0.0633], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0148, 0.0151, 0.0166, 0.0151, 0.0127, 0.0173, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:40:29,748 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134178.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:41:06,266 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3111, 4.0997, 3.9427, 4.3664, 4.1186, 3.8847, 4.3995, 3.6903], device='cuda:1'), covar=tensor([0.0495, 0.0935, 0.0476, 0.0474, 0.0757, 0.1295, 0.0565, 0.0542], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0275, 0.0200, 0.0196, 0.0184, 0.0158, 0.0290, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 13:41:18,750 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 1.953e+02 2.384e+02 3.110e+02 5.351e+02, threshold=4.768e+02, percent-clipped=0.0 2022-12-08 13:41:36,572 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 13:41:38,335 INFO [train.py:873] (1/4) Epoch 18, batch 5700, loss[loss=0.1134, simple_loss=0.1477, pruned_loss=0.03955, over 6959.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1411, pruned_loss=0.03488, over 1895054.26 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:42:21,689 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134304.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:25,188 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134308.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:26,003 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134309.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:47,524 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.380e+02 2.136e+02 2.600e+02 3.296e+02 5.182e+02, threshold=5.200e+02, percent-clipped=3.0 2022-12-08 13:43:05,227 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:06,868 INFO [train.py:873] (1/4) Epoch 18, batch 5800, loss[loss=0.1075, simple_loss=0.149, pruned_loss=0.03302, over 14285.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1409, pruned_loss=0.03362, over 1989670.01 frames. ], batch size: 31, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:43:07,821 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:16,130 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134365.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:20,490 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134370.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:44:17,489 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.136e+02 2.508e+02 3.176e+02 6.308e+02, threshold=5.017e+02, percent-clipped=2.0 2022-12-08 13:44:37,064 INFO [train.py:873] (1/4) Epoch 18, batch 5900, loss[loss=0.09984, simple_loss=0.1313, pruned_loss=0.03418, over 6915.00 frames. ], tot_loss[loss=0.1031, simple_loss=0.1404, pruned_loss=0.03289, over 2028268.86 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:45:45,582 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.002e+02 2.502e+02 3.052e+02 5.386e+02, threshold=5.003e+02, percent-clipped=2.0 2022-12-08 13:46:04,785 INFO [train.py:873] (1/4) Epoch 18, batch 6000, loss[loss=0.08111, simple_loss=0.103, pruned_loss=0.02962, over 2650.00 frames. ], tot_loss[loss=0.1032, simple_loss=0.1403, pruned_loss=0.03306, over 2012624.57 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:46:04,786 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 13:46:17,880 INFO [train.py:905] (1/4) Epoch 18, validation: loss=0.1412, simple_loss=0.1783, pruned_loss=0.05203, over 857387.00 frames. 2022-12-08 13:46:17,880 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 13:46:27,843 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8590, 4.4659, 4.3900, 4.8792, 4.5281, 4.2741, 4.8849, 4.0203], device='cuda:1'), covar=tensor([0.0362, 0.0997, 0.0425, 0.0402, 0.0786, 0.0732, 0.0495, 0.0539], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0275, 0.0200, 0.0195, 0.0183, 0.0157, 0.0288, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 13:46:58,423 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2022-12-08 13:47:20,372 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2022-12-08 13:47:26,071 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 2.089e+02 2.569e+02 3.442e+02 1.078e+03, threshold=5.138e+02, percent-clipped=4.0 2022-12-08 13:47:44,088 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:47:45,553 INFO [train.py:873] (1/4) Epoch 18, batch 6100, loss[loss=0.1135, simple_loss=0.149, pruned_loss=0.03904, over 14169.00 frames. ], tot_loss[loss=0.103, simple_loss=0.1402, pruned_loss=0.03297, over 2042706.57 frames. ], batch size: 84, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:47:49,950 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134660.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:47:54,197 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134665.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:48:25,747 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134701.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:48:42,740 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2882, 3.7505, 3.0073, 4.5147, 4.2101, 4.4055, 3.9976, 3.1029], device='cuda:1'), covar=tensor([0.0645, 0.1077, 0.2948, 0.0532, 0.0892, 0.1035, 0.0852, 0.2785], device='cuda:1'), in_proj_covar=tensor([0.0286, 0.0293, 0.0263, 0.0296, 0.0326, 0.0307, 0.0256, 0.0246], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:48:53,586 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.283e+02 2.174e+02 2.543e+02 3.201e+02 1.223e+03, threshold=5.086e+02, percent-clipped=2.0 2022-12-08 13:49:12,509 INFO [train.py:873] (1/4) Epoch 18, batch 6200, loss[loss=0.1022, simple_loss=0.1379, pruned_loss=0.03322, over 14268.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1406, pruned_loss=0.03374, over 1926680.35 frames. ], batch size: 89, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:50:20,437 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.042e+02 2.445e+02 3.055e+02 6.497e+02, threshold=4.889e+02, percent-clipped=2.0 2022-12-08 13:50:26,076 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 13:50:40,210 INFO [train.py:873] (1/4) Epoch 18, batch 6300, loss[loss=0.09791, simple_loss=0.1415, pruned_loss=0.02715, over 14332.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.14, pruned_loss=0.03289, over 1983265.01 frames. ], batch size: 66, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:50:40,349 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0957, 4.2860, 4.5417, 3.8860, 4.3261, 4.5490, 1.8660, 4.0124], device='cuda:1'), covar=tensor([0.0368, 0.0333, 0.0305, 0.0534, 0.0309, 0.0237, 0.2935, 0.0309], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0175, 0.0147, 0.0149, 0.0206, 0.0141, 0.0157, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 13:50:46,265 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134862.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:51:39,639 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134923.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:51:47,984 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.107e+02 2.599e+02 3.257e+02 7.432e+02, threshold=5.197e+02, percent-clipped=1.0 2022-12-08 13:52:03,734 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0474, 2.3791, 2.3092, 2.3927, 2.0191, 2.4570, 2.2939, 1.4081], device='cuda:1'), covar=tensor([0.1045, 0.0848, 0.0777, 0.0540, 0.1052, 0.0624, 0.1063, 0.2010], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0091, 0.0071, 0.0078, 0.0102, 0.0093, 0.0104, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 13:52:03,795 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1953, 2.2102, 3.1027, 2.3043, 3.0503, 2.9980, 2.9403, 2.5964], device='cuda:1'), covar=tensor([0.0876, 0.2883, 0.0960, 0.1925, 0.0766, 0.1125, 0.1220, 0.1783], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0309, 0.0389, 0.0301, 0.0363, 0.0323, 0.0362, 0.0297], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:52:07,216 INFO [train.py:873] (1/4) Epoch 18, batch 6400, loss[loss=0.1034, simple_loss=0.144, pruned_loss=0.03138, over 14240.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1401, pruned_loss=0.03274, over 1984321.04 frames. ], batch size: 89, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:52:07,443 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0063, 2.1594, 2.9524, 3.0630, 3.0135, 2.2617, 3.0092, 2.3781], device='cuda:1'), covar=tensor([0.0501, 0.1281, 0.0782, 0.0548, 0.0631, 0.1651, 0.0502, 0.1071], device='cuda:1'), in_proj_covar=tensor([0.0292, 0.0258, 0.0373, 0.0331, 0.0271, 0.0306, 0.0309, 0.0276], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 13:52:12,114 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134960.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:52:16,298 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134965.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:52:57,645 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:53:01,789 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135013.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:53:18,809 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.100e+02 2.565e+02 3.370e+02 5.862e+02, threshold=5.131e+02, percent-clipped=3.0 2022-12-08 13:53:18,989 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7046, 1.5557, 3.6411, 1.7157, 3.6316, 3.8175, 2.8833, 4.0622], device='cuda:1'), covar=tensor([0.0252, 0.3074, 0.0500, 0.2230, 0.0665, 0.0472, 0.0833, 0.0205], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0163, 0.0169, 0.0170, 0.0181, 0.0133, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:53:20,048 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0717, 2.1990, 2.9218, 2.3536, 2.9303, 2.8374, 2.8658, 2.5814], device='cuda:1'), covar=tensor([0.0742, 0.2273, 0.0709, 0.1410, 0.0571, 0.0933, 0.0914, 0.1410], device='cuda:1'), in_proj_covar=tensor([0.0352, 0.0308, 0.0389, 0.0300, 0.0362, 0.0323, 0.0361, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:53:38,283 INFO [train.py:873] (1/4) Epoch 18, batch 6500, loss[loss=0.0929, simple_loss=0.1368, pruned_loss=0.02451, over 14426.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1406, pruned_loss=0.03305, over 1975696.27 frames. ], batch size: 41, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:54:28,243 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7951, 1.5618, 3.9644, 3.6797, 3.7314, 4.0252, 3.3505, 4.0230], device='cuda:1'), covar=tensor([0.1676, 0.1748, 0.0139, 0.0304, 0.0279, 0.0178, 0.0306, 0.0144], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0130, 0.0168, 0.0147, 0.0143, 0.0126, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 13:54:46,810 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.038e+02 2.650e+02 3.158e+02 7.242e+02, threshold=5.300e+02, percent-clipped=3.0 2022-12-08 13:55:05,523 INFO [train.py:873] (1/4) Epoch 18, batch 6600, loss[loss=0.1193, simple_loss=0.1346, pruned_loss=0.05199, over 3848.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1407, pruned_loss=0.03346, over 2010016.55 frames. ], batch size: 100, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:55:20,649 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:56:01,507 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:56:13,671 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135232.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:56:14,313 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.150e+02 2.615e+02 3.167e+02 5.492e+02, threshold=5.230e+02, percent-clipped=1.0 2022-12-08 13:56:20,345 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 13:56:34,471 INFO [train.py:873] (1/4) Epoch 18, batch 6700, loss[loss=0.1173, simple_loss=0.1473, pruned_loss=0.04361, over 7770.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.141, pruned_loss=0.03392, over 2023708.04 frames. ], batch size: 100, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:57:41,809 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.031e+02 2.097e+02 2.445e+02 2.909e+02 5.288e+02, threshold=4.890e+02, percent-clipped=1.0 2022-12-08 13:58:00,102 INFO [train.py:873] (1/4) Epoch 18, batch 6800, loss[loss=0.09161, simple_loss=0.1313, pruned_loss=0.02597, over 13949.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1406, pruned_loss=0.0339, over 1990028.44 frames. ], batch size: 23, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 13:58:13,764 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0247, 2.0745, 2.8982, 2.3185, 2.8071, 2.7992, 2.7000, 2.4200], device='cuda:1'), covar=tensor([0.0861, 0.2934, 0.0987, 0.1763, 0.0672, 0.1102, 0.1111, 0.1615], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0310, 0.0390, 0.0300, 0.0363, 0.0322, 0.0362, 0.0297], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 13:58:41,745 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-08 13:59:07,783 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.213e+02 2.677e+02 3.514e+02 1.198e+03, threshold=5.353e+02, percent-clipped=7.0 2022-12-08 13:59:15,460 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6357, 1.7332, 1.8061, 1.3981, 1.2758, 1.6688, 1.1398, 1.6699], device='cuda:1'), covar=tensor([0.1401, 0.2277, 0.0914, 0.2325, 0.2793, 0.1131, 0.2713, 0.1244], device='cuda:1'), in_proj_covar=tensor([0.0086, 0.0102, 0.0095, 0.0099, 0.0115, 0.0092, 0.0116, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 13:59:27,058 INFO [train.py:873] (1/4) Epoch 18, batch 6900, loss[loss=0.1096, simple_loss=0.1385, pruned_loss=0.04036, over 4962.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.1408, pruned_loss=0.03424, over 1912393.06 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 13:59:51,271 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 14:00:21,774 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135518.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:00:29,530 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135527.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:00:35,294 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.207e+02 2.585e+02 3.410e+02 8.303e+02, threshold=5.169e+02, percent-clipped=6.0 2022-12-08 14:00:53,012 INFO [train.py:873] (1/4) Epoch 18, batch 7000, loss[loss=0.1128, simple_loss=0.1478, pruned_loss=0.03883, over 14378.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1406, pruned_loss=0.03365, over 1953123.88 frames. ], batch size: 53, lr: 4.31e-03, grad_scale: 4.0 2022-12-08 14:01:03,617 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135566.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:01:27,342 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 14:01:36,608 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9265, 4.6187, 4.4204, 4.9626, 4.5975, 4.3337, 4.9417, 4.1914], device='cuda:1'), covar=tensor([0.0346, 0.0933, 0.0391, 0.0412, 0.0760, 0.0515, 0.0558, 0.0471], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0280, 0.0203, 0.0200, 0.0186, 0.0161, 0.0291, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:01:59,466 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:02:02,411 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 2.039e+02 2.453e+02 3.005e+02 1.669e+03, threshold=4.906e+02, percent-clipped=2.0 2022-12-08 14:02:02,605 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8587, 1.5667, 1.8952, 1.6145, 1.9038, 1.7237, 1.5750, 1.7935], device='cuda:1'), covar=tensor([0.0701, 0.1650, 0.0675, 0.0690, 0.0649, 0.0880, 0.0451, 0.0529], device='cuda:1'), in_proj_covar=tensor([0.0353, 0.0310, 0.0392, 0.0300, 0.0364, 0.0323, 0.0362, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:02:05,255 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5439, 1.8696, 1.9714, 1.9755, 1.8164, 2.0457, 1.7358, 1.4310], device='cuda:1'), covar=tensor([0.1011, 0.1281, 0.0730, 0.0697, 0.1295, 0.0837, 0.1558, 0.1907], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0090, 0.0071, 0.0077, 0.0102, 0.0092, 0.0103, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:02:20,527 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.07 vs. limit=5.0 2022-12-08 14:02:21,584 INFO [train.py:873] (1/4) Epoch 18, batch 7100, loss[loss=0.1337, simple_loss=0.1535, pruned_loss=0.05697, over 7774.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1409, pruned_loss=0.0338, over 1975903.69 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 4.0 2022-12-08 14:02:53,360 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:03:24,697 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-08 14:03:30,724 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.202e+02 1.933e+02 2.466e+02 3.185e+02 5.052e+02, threshold=4.932e+02, percent-clipped=2.0 2022-12-08 14:03:48,867 INFO [train.py:873] (1/4) Epoch 18, batch 7200, loss[loss=0.1236, simple_loss=0.1537, pruned_loss=0.04671, over 10352.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.141, pruned_loss=0.03375, over 2005500.71 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 14:04:53,006 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135827.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:04:58,957 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 2.075e+02 2.523e+02 3.199e+02 6.288e+02, threshold=5.045e+02, percent-clipped=2.0 2022-12-08 14:05:18,066 INFO [train.py:873] (1/4) Epoch 18, batch 7300, loss[loss=0.09876, simple_loss=0.1424, pruned_loss=0.02757, over 14355.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1402, pruned_loss=0.03352, over 1973268.87 frames. ], batch size: 55, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 14:05:35,233 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:06:27,399 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.731e+01 2.137e+02 2.530e+02 3.235e+02 6.685e+02, threshold=5.060e+02, percent-clipped=2.0 2022-12-08 14:06:42,021 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-08 14:06:45,611 INFO [train.py:873] (1/4) Epoch 18, batch 7400, loss[loss=0.1308, simple_loss=0.1556, pruned_loss=0.05305, over 11994.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1401, pruned_loss=0.03369, over 1961985.82 frames. ], batch size: 100, lr: 4.30e-03, grad_scale: 8.0 2022-12-08 14:06:56,114 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8730, 2.6610, 2.4719, 2.5917, 2.7917, 2.8132, 2.8259, 2.8322], device='cuda:1'), covar=tensor([0.1014, 0.0820, 0.2266, 0.2569, 0.1031, 0.1088, 0.1127, 0.0904], device='cuda:1'), in_proj_covar=tensor([0.0394, 0.0276, 0.0451, 0.0572, 0.0351, 0.0455, 0.0385, 0.0399], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:07:09,846 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135982.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:13,160 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135986.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:20,220 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135994.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:23,191 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.47 vs. limit=2.0 2022-12-08 14:07:36,725 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5500, 3.8534, 3.8325, 3.5385, 2.7180, 3.8614, 3.6529, 2.0505], device='cuda:1'), covar=tensor([0.1215, 0.0757, 0.0516, 0.1677, 0.0951, 0.0493, 0.1065, 0.1888], device='cuda:1'), in_proj_covar=tensor([0.0139, 0.0091, 0.0071, 0.0078, 0.0102, 0.0093, 0.0103, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:07:55,412 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.076e+02 2.483e+02 2.988e+02 6.746e+02, threshold=4.966e+02, percent-clipped=1.0 2022-12-08 14:08:04,508 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:08:14,774 INFO [train.py:873] (1/4) Epoch 18, batch 7500, loss[loss=0.1032, simple_loss=0.1427, pruned_loss=0.03191, over 14192.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1403, pruned_loss=0.03401, over 1931558.09 frames. ], batch size: 69, lr: 4.30e-03, grad_scale: 8.0 2022-12-08 14:08:14,907 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:08:50,042 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4693, 2.5059, 2.4502, 2.3663, 2.3688, 1.5058, 2.0831, 2.5185], device='cuda:1'), covar=tensor([0.1304, 0.0612, 0.0450, 0.0654, 0.0993, 0.0739, 0.0963, 0.0720], device='cuda:1'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0038, 0.0040], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 14:09:45,354 INFO [train.py:873] (1/4) Epoch 19, batch 0, loss[loss=0.1619, simple_loss=0.1779, pruned_loss=0.07298, over 8618.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.1779, pruned_loss=0.07298, over 8618.00 frames. ], batch size: 100, lr: 4.19e-03, grad_scale: 8.0 2022-12-08 14:09:45,354 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 14:09:53,121 INFO [train.py:905] (1/4) Epoch 19, validation: loss=0.1445, simple_loss=0.1825, pruned_loss=0.05324, over 857387.00 frames. 2022-12-08 14:09:53,122 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 14:09:58,595 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.01 vs. limit=5.0 2022-12-08 14:10:08,830 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.063e+01 1.495e+02 2.536e+02 3.389e+02 8.096e+02, threshold=5.072e+02, percent-clipped=9.0 2022-12-08 14:10:16,277 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:10:31,692 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4070, 2.0759, 2.6838, 1.5817, 1.8917, 2.5463, 1.4948, 2.4506], device='cuda:1'), covar=tensor([0.0889, 0.1512, 0.0691, 0.2576, 0.2149, 0.0764, 0.3044, 0.0789], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0103, 0.0096, 0.0100, 0.0115, 0.0093, 0.0117, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 14:10:34,050 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6614, 3.6137, 3.4082, 3.7204, 3.2979, 3.1421, 3.7533, 3.5469], device='cuda:1'), covar=tensor([0.0694, 0.0937, 0.1049, 0.0647, 0.1263, 0.1001, 0.0683, 0.0945], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0148, 0.0150, 0.0166, 0.0152, 0.0126, 0.0175, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:10:59,121 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 14:11:10,061 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136203.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:11:22,946 INFO [train.py:873] (1/4) Epoch 19, batch 100, loss[loss=0.09626, simple_loss=0.1277, pruned_loss=0.03242, over 3847.00 frames. ], tot_loss[loss=0.104, simple_loss=0.142, pruned_loss=0.03302, over 843573.31 frames. ], batch size: 100, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:11:23,486 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0212, 2.1454, 3.9519, 2.7496, 3.8467, 1.9710, 3.0415, 3.9122], device='cuda:1'), covar=tensor([0.0680, 0.3722, 0.0554, 0.4970, 0.0682, 0.3389, 0.1372, 0.0498], device='cuda:1'), in_proj_covar=tensor([0.0254, 0.0199, 0.0219, 0.0268, 0.0238, 0.0203, 0.0201, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 14:11:23,848 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-08 14:11:36,009 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7092, 3.7855, 3.9848, 3.5690, 3.8681, 3.8792, 1.4909, 3.6376], device='cuda:1'), covar=tensor([0.0364, 0.0369, 0.0324, 0.0452, 0.0298, 0.0345, 0.3088, 0.0280], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0176, 0.0148, 0.0150, 0.0208, 0.0141, 0.0158, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:11:37,511 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.463e+02 2.960e+02 3.618e+02 1.073e+03, threshold=5.920e+02, percent-clipped=4.0 2022-12-08 14:12:23,025 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:12:43,769 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=6.58 vs. limit=5.0 2022-12-08 14:12:50,168 INFO [train.py:873] (1/4) Epoch 19, batch 200, loss[loss=0.1208, simple_loss=0.1202, pruned_loss=0.06066, over 1273.00 frames. ], tot_loss[loss=0.1031, simple_loss=0.1408, pruned_loss=0.03268, over 1244709.58 frames. ], batch size: 100, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:13:05,476 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136334.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:05,830 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 14:13:06,200 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.129e+02 2.646e+02 3.149e+02 5.242e+02, threshold=5.292e+02, percent-clipped=0.0 2022-12-08 14:13:06,391 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:09,309 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136338.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:16,923 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7022, 2.4252, 3.1049, 2.0859, 2.0529, 2.7065, 1.5522, 2.8435], device='cuda:1'), covar=tensor([0.0928, 0.1110, 0.0618, 0.1755, 0.1977, 0.0927, 0.2786, 0.0913], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0102, 0.0096, 0.0100, 0.0115, 0.0092, 0.0117, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 14:13:19,503 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:59,653 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:14:18,704 INFO [train.py:873] (1/4) Epoch 19, batch 300, loss[loss=0.1118, simple_loss=0.1439, pruned_loss=0.03987, over 14244.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.1416, pruned_loss=0.03368, over 1581733.89 frames. ], batch size: 80, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:14:33,862 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 1.988e+02 2.460e+02 2.943e+02 5.876e+02, threshold=4.921e+02, percent-clipped=3.0 2022-12-08 14:14:40,004 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0274, 4.6388, 4.5186, 5.0527, 4.7588, 4.4606, 5.0452, 4.2192], device='cuda:1'), covar=tensor([0.0338, 0.0856, 0.0401, 0.0367, 0.0664, 0.0598, 0.0447, 0.0493], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0275, 0.0201, 0.0197, 0.0185, 0.0159, 0.0287, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:14:50,906 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2634, 2.1200, 3.2722, 3.4012, 3.2101, 2.2139, 3.3227, 2.4830], device='cuda:1'), covar=tensor([0.0576, 0.1487, 0.0876, 0.0614, 0.0646, 0.2068, 0.0542, 0.1240], device='cuda:1'), in_proj_covar=tensor([0.0298, 0.0264, 0.0380, 0.0336, 0.0275, 0.0311, 0.0314, 0.0282], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:14:56,932 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2057, 2.3632, 5.0912, 4.6381, 4.4013, 5.2105, 4.9186, 5.1946], device='cuda:1'), covar=tensor([0.1605, 0.1374, 0.0091, 0.0192, 0.0219, 0.0126, 0.0157, 0.0102], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0171, 0.0148, 0.0143, 0.0128, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 14:14:57,781 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0876, 4.1877, 4.4122, 3.7530, 4.2739, 4.4579, 1.6990, 4.0081], device='cuda:1'), covar=tensor([0.0324, 0.0342, 0.0343, 0.0604, 0.0313, 0.0231, 0.3071, 0.0288], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0176, 0.0147, 0.0149, 0.0208, 0.0141, 0.0158, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:15:02,338 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 14:15:17,142 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2022-12-08 14:15:19,986 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2971, 3.0865, 3.0553, 3.3133, 3.1044, 3.2532, 3.3609, 2.8202], device='cuda:1'), covar=tensor([0.0465, 0.0929, 0.0548, 0.0510, 0.0811, 0.0434, 0.0537, 0.0609], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0274, 0.0200, 0.0196, 0.0184, 0.0159, 0.0286, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:15:29,507 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136498.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:15:33,608 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 14:15:45,665 INFO [train.py:873] (1/4) Epoch 19, batch 400, loss[loss=0.1065, simple_loss=0.131, pruned_loss=0.04096, over 3876.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1413, pruned_loss=0.0336, over 1792056.42 frames. ], batch size: 100, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:16:01,299 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.033e+02 2.479e+02 3.164e+02 1.103e+03, threshold=4.959e+02, percent-clipped=2.0 2022-12-08 14:16:02,229 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-08 14:17:14,357 INFO [train.py:873] (1/4) Epoch 19, batch 500, loss[loss=0.107, simple_loss=0.1461, pruned_loss=0.0339, over 14189.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1407, pruned_loss=0.03328, over 1890233.71 frames. ], batch size: 84, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:17:30,014 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 1.937e+02 2.457e+02 2.957e+02 6.738e+02, threshold=4.914e+02, percent-clipped=4.0 2022-12-08 14:17:32,462 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:36,757 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:41,093 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.8114, 5.2668, 5.2728, 5.7672, 5.4772, 4.7251, 5.7410, 4.7822], device='cuda:1'), covar=tensor([0.0347, 0.0942, 0.0331, 0.0353, 0.0639, 0.0377, 0.0469, 0.0455], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0274, 0.0200, 0.0197, 0.0184, 0.0159, 0.0286, 0.0168], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:17:42,851 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:42,953 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1374, 2.1316, 3.0741, 3.2119, 3.0695, 2.1989, 3.0837, 2.4679], device='cuda:1'), covar=tensor([0.0589, 0.1326, 0.0924, 0.0669, 0.0669, 0.1911, 0.0565, 0.1147], device='cuda:1'), in_proj_covar=tensor([0.0297, 0.0263, 0.0380, 0.0335, 0.0275, 0.0310, 0.0313, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:18:13,551 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:17,691 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:24,490 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136698.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:24,935 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 14:18:29,975 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:40,751 INFO [train.py:873] (1/4) Epoch 19, batch 600, loss[loss=0.1334, simple_loss=0.1643, pruned_loss=0.05121, over 14301.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1406, pruned_loss=0.03314, over 1921138.81 frames. ], batch size: 35, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:18:57,079 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 2.028e+02 2.404e+02 2.918e+02 4.938e+02, threshold=4.808e+02, percent-clipped=1.0 2022-12-08 14:19:51,263 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136798.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:19:58,228 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:20:02,309 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-08 14:20:08,082 INFO [train.py:873] (1/4) Epoch 19, batch 700, loss[loss=0.1312, simple_loss=0.1586, pruned_loss=0.0519, over 9467.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.14, pruned_loss=0.03234, over 1999953.69 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:20:24,140 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.847e+01 1.962e+02 2.462e+02 3.266e+02 7.518e+02, threshold=4.924e+02, percent-clipped=4.0 2022-12-08 14:20:32,583 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136846.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:20:33,253 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 14:20:51,839 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136867.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:21:35,621 INFO [train.py:873] (1/4) Epoch 19, batch 800, loss[loss=0.09479, simple_loss=0.1262, pruned_loss=0.03166, over 6022.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1408, pruned_loss=0.03332, over 2000880.00 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:21:52,113 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.241e+02 2.727e+02 3.173e+02 1.328e+03, threshold=5.453e+02, percent-clipped=3.0 2022-12-08 14:22:15,159 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0046, 2.9228, 2.2396, 3.0525, 2.8936, 2.9608, 2.5292, 2.2913], device='cuda:1'), covar=tensor([0.0948, 0.1176, 0.2986, 0.0862, 0.1086, 0.0989, 0.1519, 0.2643], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0286, 0.0257, 0.0288, 0.0320, 0.0301, 0.0253, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:22:40,689 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136991.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:22:47,228 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136999.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:23:00,018 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4662, 3.5477, 3.7507, 3.4270, 3.6250, 3.4479, 1.4872, 3.4067], device='cuda:1'), covar=tensor([0.0395, 0.0389, 0.0330, 0.0466, 0.0303, 0.0523, 0.3228, 0.0329], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0149, 0.0210, 0.0142, 0.0158, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:23:03,205 INFO [train.py:873] (1/4) Epoch 19, batch 900, loss[loss=0.162, simple_loss=0.1527, pruned_loss=0.0856, over 1179.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1415, pruned_loss=0.03394, over 1950734.14 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:23:20,504 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.006e+02 2.396e+02 3.046e+02 6.535e+02, threshold=4.793e+02, percent-clipped=3.0 2022-12-08 14:23:22,041 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137039.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:23:23,785 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2760, 4.3551, 4.5803, 4.0552, 4.4061, 4.5723, 1.7155, 4.1547], device='cuda:1'), covar=tensor([0.0334, 0.0292, 0.0310, 0.0424, 0.0288, 0.0205, 0.3085, 0.0278], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0176, 0.0148, 0.0149, 0.0209, 0.0142, 0.0158, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:24:29,763 INFO [train.py:873] (1/4) Epoch 19, batch 1000, loss[loss=0.1159, simple_loss=0.1386, pruned_loss=0.0466, over 5958.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1417, pruned_loss=0.03441, over 1925933.58 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:24:47,893 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.038e+02 2.060e+02 2.475e+02 3.449e+02 7.129e+02, threshold=4.949e+02, percent-clipped=6.0 2022-12-08 14:24:54,227 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 14:25:06,121 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8181, 1.5203, 2.0278, 1.6592, 1.9105, 1.4429, 1.6740, 1.9253], device='cuda:1'), covar=tensor([0.3126, 0.2936, 0.0625, 0.1843, 0.1489, 0.1382, 0.1204, 0.1164], device='cuda:1'), in_proj_covar=tensor([0.0255, 0.0198, 0.0218, 0.0271, 0.0239, 0.0201, 0.0201, 0.0219], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 14:25:09,575 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=137162.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:25:57,946 INFO [train.py:873] (1/4) Epoch 19, batch 1100, loss[loss=0.1098, simple_loss=0.1483, pruned_loss=0.03563, over 6945.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.1419, pruned_loss=0.035, over 1889288.12 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:26:03,511 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6493, 1.3028, 3.7537, 1.6238, 3.6771, 3.8694, 2.8124, 4.0482], device='cuda:1'), covar=tensor([0.0292, 0.3536, 0.0450, 0.2497, 0.0698, 0.0412, 0.0788, 0.0238], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0161, 0.0168, 0.0168, 0.0179, 0.0132, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:26:15,916 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 1.998e+02 2.599e+02 3.181e+02 8.198e+02, threshold=5.197e+02, percent-clipped=8.0 2022-12-08 14:26:19,989 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-08 14:26:55,114 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7543, 2.7743, 2.6135, 2.8998, 2.4661, 2.6626, 2.8501, 2.7745], device='cuda:1'), covar=tensor([0.0882, 0.1247, 0.1026, 0.0747, 0.1145, 0.0801, 0.0898, 0.0899], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0149, 0.0150, 0.0166, 0.0153, 0.0128, 0.0175, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:27:10,340 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:27:13,199 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 14:27:25,930 INFO [train.py:873] (1/4) Epoch 19, batch 1200, loss[loss=0.09319, simple_loss=0.1202, pruned_loss=0.03311, over 5939.00 frames. ], tot_loss[loss=0.1052, simple_loss=0.1417, pruned_loss=0.03442, over 1941413.97 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:27:43,343 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.257e+02 2.745e+02 3.542e+02 9.944e+02, threshold=5.490e+02, percent-clipped=8.0 2022-12-08 14:27:51,868 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:28:12,238 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3831, 3.8041, 2.9176, 4.6861, 4.3193, 4.4877, 3.8961, 3.2143], device='cuda:1'), covar=tensor([0.0677, 0.1128, 0.3431, 0.0505, 0.0686, 0.1163, 0.1096, 0.3359], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0286, 0.0258, 0.0290, 0.0319, 0.0302, 0.0254, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:28:46,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0590, 4.7552, 4.4413, 4.6561, 4.6828, 4.9396, 5.0299, 5.0256], device='cuda:1'), covar=tensor([0.0791, 0.0491, 0.2291, 0.2684, 0.0799, 0.0830, 0.0767, 0.0820], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0281, 0.0459, 0.0578, 0.0356, 0.0464, 0.0396, 0.0407], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:28:54,280 INFO [train.py:873] (1/4) Epoch 19, batch 1300, loss[loss=0.1323, simple_loss=0.1583, pruned_loss=0.05316, over 12006.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.141, pruned_loss=0.03374, over 1980364.09 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:29:12,653 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.901e+02 2.314e+02 2.897e+02 7.672e+02, threshold=4.628e+02, percent-clipped=3.0 2022-12-08 14:29:33,979 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:30:00,703 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4788, 3.8545, 2.8616, 4.7132, 4.2623, 4.5102, 3.9840, 3.3585], device='cuda:1'), covar=tensor([0.0629, 0.1107, 0.3204, 0.0698, 0.0944, 0.1281, 0.1066, 0.2587], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0286, 0.0258, 0.0290, 0.0320, 0.0302, 0.0253, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:30:15,550 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:30:21,589 INFO [train.py:873] (1/4) Epoch 19, batch 1400, loss[loss=0.1437, simple_loss=0.135, pruned_loss=0.07621, over 1219.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1406, pruned_loss=0.03373, over 1981424.13 frames. ], batch size: 100, lr: 4.16e-03, grad_scale: 4.0 2022-12-08 14:30:40,256 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.087e+02 2.410e+02 2.920e+02 5.806e+02, threshold=4.820e+02, percent-clipped=3.0 2022-12-08 14:30:47,687 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4700, 2.5036, 1.7346, 2.5476, 2.4209, 2.2400, 2.1453, 2.2312], device='cuda:1'), covar=tensor([0.0313, 0.0866, 0.0702, 0.0213, 0.0257, 0.0395, 0.0457, 0.0453], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 14:30:49,306 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5965, 1.4965, 1.5815, 1.6721, 1.6779, 1.1341, 1.4965, 1.4976], device='cuda:1'), covar=tensor([0.0792, 0.0655, 0.0882, 0.0693, 0.0663, 0.0886, 0.0781, 0.0603], device='cuda:1'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0050, 0.0038, 0.0040], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 14:31:07,346 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8209, 1.5380, 1.7725, 1.9411, 1.4028, 1.7224, 1.6590, 1.7883], device='cuda:1'), covar=tensor([0.0299, 0.0486, 0.0333, 0.0275, 0.0494, 0.0565, 0.0349, 0.0312], device='cuda:1'), in_proj_covar=tensor([0.0298, 0.0264, 0.0381, 0.0336, 0.0277, 0.0311, 0.0315, 0.0281], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:31:13,390 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4182, 3.1664, 2.5160, 3.5215, 3.3699, 3.3830, 3.0399, 2.4725], device='cuda:1'), covar=tensor([0.0835, 0.1371, 0.2897, 0.0633, 0.0962, 0.0993, 0.1238, 0.2853], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0288, 0.0260, 0.0292, 0.0323, 0.0304, 0.0254, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:31:49,006 INFO [train.py:873] (1/4) Epoch 19, batch 1500, loss[loss=0.114, simple_loss=0.15, pruned_loss=0.03904, over 14223.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1405, pruned_loss=0.03353, over 1980990.64 frames. ], batch size: 94, lr: 4.16e-03, grad_scale: 4.0 2022-12-08 14:32:01,261 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 14:32:07,872 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 1.988e+02 2.718e+02 3.293e+02 9.468e+02, threshold=5.435e+02, percent-clipped=6.0 2022-12-08 14:32:17,550 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3752, 2.9797, 2.9389, 1.8980, 2.8509, 3.0996, 3.3538, 2.5397], device='cuda:1'), covar=tensor([0.0699, 0.0987, 0.0989, 0.1669, 0.1095, 0.0684, 0.0764, 0.1367], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0170, 0.0142, 0.0126, 0.0146, 0.0156, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:32:35,525 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2414, 3.3085, 3.3679, 3.1725, 3.3172, 3.0591, 1.5377, 3.0725], device='cuda:1'), covar=tensor([0.0523, 0.0415, 0.0474, 0.0536, 0.0492, 0.0829, 0.3353, 0.0480], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0177, 0.0149, 0.0150, 0.0211, 0.0144, 0.0159, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:33:18,036 INFO [train.py:873] (1/4) Epoch 19, batch 1600, loss[loss=0.18, simple_loss=0.1667, pruned_loss=0.09663, over 1341.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.1408, pruned_loss=0.03397, over 1938174.56 frames. ], batch size: 100, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:33:36,638 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.099e+02 2.503e+02 3.045e+02 5.451e+02, threshold=5.005e+02, percent-clipped=1.0 2022-12-08 14:33:59,637 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0843, 2.1170, 2.1476, 2.4256, 2.1143, 1.4345, 1.9989, 2.1609], device='cuda:1'), covar=tensor([0.0741, 0.0699, 0.0840, 0.0390, 0.0631, 0.0693, 0.0662, 0.0455], device='cuda:1'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0038, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 14:34:24,383 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8200, 3.2342, 2.9648, 3.2417, 2.4162, 3.3062, 3.0190, 1.7630], device='cuda:1'), covar=tensor([0.1098, 0.0868, 0.0861, 0.0481, 0.0913, 0.0477, 0.1004, 0.1810], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0091, 0.0070, 0.0077, 0.0101, 0.0092, 0.0102, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:34:46,399 INFO [train.py:873] (1/4) Epoch 19, batch 1700, loss[loss=0.1476, simple_loss=0.1411, pruned_loss=0.07706, over 2607.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.1411, pruned_loss=0.03425, over 1953393.02 frames. ], batch size: 100, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:34:53,080 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6017, 2.5239, 1.9703, 2.6606, 2.4910, 2.4651, 2.2282, 2.0931], device='cuda:1'), covar=tensor([0.1008, 0.1137, 0.2331, 0.0922, 0.1242, 0.1126, 0.1651, 0.2229], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0289, 0.0261, 0.0292, 0.0324, 0.0305, 0.0256, 0.0244], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:35:03,831 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0125, 3.4850, 2.7337, 4.1518, 3.9709, 3.9895, 3.5396, 2.9638], device='cuda:1'), covar=tensor([0.0771, 0.1189, 0.3023, 0.0576, 0.0877, 0.1105, 0.1045, 0.2615], device='cuda:1'), in_proj_covar=tensor([0.0283, 0.0288, 0.0260, 0.0292, 0.0324, 0.0304, 0.0256, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:35:05,299 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.077e+02 2.557e+02 3.257e+02 5.458e+02, threshold=5.115e+02, percent-clipped=1.0 2022-12-08 14:35:47,398 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8860, 1.8847, 2.1003, 1.7386, 1.7094, 1.7391, 1.7325, 1.3482], device='cuda:1'), covar=tensor([0.0195, 0.0286, 0.0148, 0.0344, 0.0284, 0.0302, 0.0289, 0.0449], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 14:35:53,039 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:35:55,475 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137895.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:36:14,637 INFO [train.py:873] (1/4) Epoch 19, batch 1800, loss[loss=0.0673, simple_loss=0.1078, pruned_loss=0.01342, over 10776.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1413, pruned_loss=0.03422, over 1955645.06 frames. ], batch size: 13, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:36:33,146 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 2.138e+02 2.501e+02 2.995e+02 6.998e+02, threshold=5.002e+02, percent-clipped=2.0 2022-12-08 14:36:46,406 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-08 14:36:46,878 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137953.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:36:49,330 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137956.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:36:52,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4508, 4.5184, 4.8437, 4.1350, 4.6612, 4.8718, 1.8475, 4.3494], device='cuda:1'), covar=tensor([0.0358, 0.0353, 0.0352, 0.0537, 0.0314, 0.0213, 0.3265, 0.0338], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0175, 0.0148, 0.0149, 0.0209, 0.0143, 0.0159, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:37:42,664 INFO [train.py:873] (1/4) Epoch 19, batch 1900, loss[loss=0.08614, simple_loss=0.1319, pruned_loss=0.02019, over 14036.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1406, pruned_loss=0.03349, over 1941369.71 frames. ], batch size: 26, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:37:53,046 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2022-12-08 14:38:01,167 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.140e+02 2.687e+02 3.163e+02 5.638e+02, threshold=5.373e+02, percent-clipped=1.0 2022-12-08 14:38:05,989 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2022-12-08 14:38:14,288 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2022-12-08 14:38:27,759 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3672, 3.2708, 3.0713, 3.4323, 3.0524, 3.0259, 3.4152, 3.3124], device='cuda:1'), covar=tensor([0.0664, 0.1085, 0.1079, 0.0706, 0.1077, 0.0748, 0.0750, 0.0858], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0147, 0.0150, 0.0165, 0.0150, 0.0127, 0.0172, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:38:51,335 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.37 vs. limit=5.0 2022-12-08 14:39:09,900 INFO [train.py:873] (1/4) Epoch 19, batch 2000, loss[loss=0.1012, simple_loss=0.1463, pruned_loss=0.02811, over 14015.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1406, pruned_loss=0.03392, over 1921580.85 frames. ], batch size: 22, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:39:28,120 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 1.998e+02 2.546e+02 3.326e+02 7.451e+02, threshold=5.092e+02, percent-clipped=3.0 2022-12-08 14:39:34,237 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2022-12-08 14:40:03,794 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8954, 4.0367, 4.2271, 3.6665, 4.0707, 4.2260, 1.6556, 3.8831], device='cuda:1'), covar=tensor([0.0378, 0.0365, 0.0341, 0.0543, 0.0309, 0.0303, 0.3175, 0.0333], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0176, 0.0147, 0.0149, 0.0209, 0.0142, 0.0158, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:40:27,458 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7889, 3.4902, 3.1732, 2.3555, 3.2608, 3.4535, 3.9844, 2.8345], device='cuda:1'), covar=tensor([0.0541, 0.0867, 0.0880, 0.1263, 0.0788, 0.0602, 0.0548, 0.1175], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0167, 0.0138, 0.0124, 0.0143, 0.0154, 0.0137, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:40:37,634 INFO [train.py:873] (1/4) Epoch 19, batch 2100, loss[loss=0.09443, simple_loss=0.132, pruned_loss=0.02842, over 14109.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1396, pruned_loss=0.03295, over 1988252.16 frames. ], batch size: 29, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:40:56,378 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 1.995e+02 2.366e+02 2.965e+02 6.942e+02, threshold=4.733e+02, percent-clipped=2.0 2022-12-08 14:41:04,849 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138248.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:41:07,339 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138251.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:41:24,557 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138270.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:42:05,013 INFO [train.py:873] (1/4) Epoch 19, batch 2200, loss[loss=0.122, simple_loss=0.1274, pruned_loss=0.0583, over 1284.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1406, pruned_loss=0.03358, over 1988502.60 frames. ], batch size: 100, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:42:09,670 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138322.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:42:17,333 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138331.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:42:22,959 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.085e+02 2.687e+02 3.331e+02 7.282e+02, threshold=5.373e+02, percent-clipped=5.0 2022-12-08 14:43:02,758 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:43:32,132 INFO [train.py:873] (1/4) Epoch 19, batch 2300, loss[loss=0.0966, simple_loss=0.1406, pruned_loss=0.02629, over 14287.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.1402, pruned_loss=0.03338, over 1980641.96 frames. ], batch size: 44, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:43:39,048 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2793, 1.8439, 2.1139, 1.9803, 2.1646, 2.0522, 1.9106, 1.6286], device='cuda:1'), covar=tensor([0.0225, 0.0608, 0.0286, 0.0408, 0.0303, 0.0398, 0.0334, 0.0456], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 14:43:50,853 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.988e+01 2.098e+02 2.453e+02 3.240e+02 6.762e+02, threshold=4.906e+02, percent-clipped=1.0 2022-12-08 14:43:53,213 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3141, 2.2117, 2.5781, 1.6819, 1.8181, 2.3522, 1.4584, 2.3407], device='cuda:1'), covar=tensor([0.0897, 0.1368, 0.0705, 0.2039, 0.2103, 0.0778, 0.3010, 0.0898], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0104, 0.0097, 0.0101, 0.0117, 0.0093, 0.0118, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 14:43:56,671 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0549, 2.3454, 3.9062, 4.2367, 4.0581, 2.2172, 4.3027, 2.8629], device='cuda:1'), covar=tensor([0.0540, 0.1566, 0.1120, 0.0521, 0.0578, 0.2595, 0.0404, 0.1461], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0263, 0.0376, 0.0333, 0.0274, 0.0308, 0.0313, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:44:28,673 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1030, 2.0411, 2.2032, 1.9265, 1.8017, 1.2807, 2.0286, 2.2766], device='cuda:1'), covar=tensor([0.0782, 0.0712, 0.0627, 0.1207, 0.1515, 0.0866, 0.0971, 0.0582], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0036, 0.0042, 0.0035, 0.0037, 0.0051, 0.0038, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 14:44:52,217 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 14:45:01,195 INFO [train.py:873] (1/4) Epoch 19, batch 2400, loss[loss=0.09049, simple_loss=0.1328, pruned_loss=0.02409, over 13971.00 frames. ], tot_loss[loss=0.1031, simple_loss=0.14, pruned_loss=0.03311, over 1961123.97 frames. ], batch size: 19, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:45:08,352 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2488, 3.6718, 3.3625, 3.3929, 2.5464, 3.5348, 3.4198, 1.8732], device='cuda:1'), covar=tensor([0.1157, 0.0530, 0.1293, 0.0905, 0.0948, 0.0583, 0.0806, 0.1847], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0091, 0.0070, 0.0077, 0.0101, 0.0092, 0.0101, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:45:18,699 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.048e+02 2.033e+02 2.636e+02 3.505e+02 6.447e+02, threshold=5.271e+02, percent-clipped=6.0 2022-12-08 14:45:20,552 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4393, 1.4584, 1.5207, 1.6058, 1.5445, 1.0728, 1.4777, 1.4208], device='cuda:1'), covar=tensor([0.0665, 0.0843, 0.0716, 0.0540, 0.0916, 0.0908, 0.0583, 0.0702], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0036, 0.0042, 0.0035, 0.0037, 0.0051, 0.0038, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 14:45:28,134 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:45:30,931 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:09,971 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138596.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:12,433 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138599.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:28,221 INFO [train.py:873] (1/4) Epoch 19, batch 2500, loss[loss=0.08459, simple_loss=0.1239, pruned_loss=0.02266, over 14619.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1399, pruned_loss=0.03338, over 1954563.33 frames. ], batch size: 21, lr: 4.15e-03, grad_scale: 4.0 2022-12-08 14:46:36,653 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:47,812 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.139e+02 2.554e+02 3.094e+02 5.985e+02, threshold=5.107e+02, percent-clipped=2.0 2022-12-08 14:46:54,591 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-08 14:46:57,709 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 14:47:21,866 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:24,787 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138681.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:29,032 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:38,342 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138697.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:38,664 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 14:47:55,442 INFO [train.py:873] (1/4) Epoch 19, batch 2600, loss[loss=0.1055, simple_loss=0.1438, pruned_loss=0.03359, over 14230.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.14, pruned_loss=0.03331, over 1950033.21 frames. ], batch size: 69, lr: 4.15e-03, grad_scale: 2.0 2022-12-08 14:48:15,369 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.052e+02 2.043e+02 2.512e+02 3.155e+02 8.164e+02, threshold=5.025e+02, percent-clipped=4.0 2022-12-08 14:48:17,570 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138742.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:21,041 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3645, 3.0675, 2.9385, 1.9471, 2.8286, 3.0952, 3.4018, 2.6746], device='cuda:1'), covar=tensor([0.0571, 0.0906, 0.0903, 0.1429, 0.0858, 0.0805, 0.0755, 0.1184], device='cuda:1'), in_proj_covar=tensor([0.0153, 0.0167, 0.0139, 0.0124, 0.0143, 0.0154, 0.0137, 0.0141], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:48:21,885 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138747.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:22,589 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1769, 3.1109, 2.9747, 3.2676, 2.8380, 2.8823, 3.2472, 3.1109], device='cuda:1'), covar=tensor([0.0661, 0.1006, 0.0861, 0.0627, 0.1150, 0.0785, 0.0731, 0.0908], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0148, 0.0150, 0.0166, 0.0152, 0.0126, 0.0174, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 14:48:31,391 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138758.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:45,359 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4457, 5.2299, 4.8808, 5.0364, 5.1446, 5.3941, 5.4271, 5.4571], device='cuda:1'), covar=tensor([0.0622, 0.0350, 0.1958, 0.2340, 0.0594, 0.0676, 0.0650, 0.0626], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0276, 0.0452, 0.0571, 0.0359, 0.0463, 0.0391, 0.0404], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:48:54,087 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9918, 4.0982, 4.3581, 3.8387, 4.2169, 4.3920, 1.7448, 3.9928], device='cuda:1'), covar=tensor([0.0369, 0.0374, 0.0360, 0.0479, 0.0312, 0.0254, 0.2980, 0.0292], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0150, 0.0209, 0.0143, 0.0158, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:49:23,531 INFO [train.py:873] (1/4) Epoch 19, batch 2700, loss[loss=0.1042, simple_loss=0.1417, pruned_loss=0.03336, over 14270.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.1401, pruned_loss=0.03324, over 1969844.67 frames. ], batch size: 76, lr: 4.14e-03, grad_scale: 2.0 2022-12-08 14:49:43,635 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.157e+02 2.518e+02 3.000e+02 5.718e+02, threshold=5.036e+02, percent-clipped=2.0 2022-12-08 14:50:07,540 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2033, 1.0731, 1.2361, 1.0198, 0.9620, 0.7208, 0.8845, 0.8049], device='cuda:1'), covar=tensor([0.0218, 0.0220, 0.0187, 0.0215, 0.0222, 0.0405, 0.0287, 0.0360], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 14:50:20,471 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 14:50:44,427 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138909.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:50:51,208 INFO [train.py:873] (1/4) Epoch 19, batch 2800, loss[loss=0.1158, simple_loss=0.1485, pruned_loss=0.04151, over 14420.00 frames. ], tot_loss[loss=0.1032, simple_loss=0.1401, pruned_loss=0.03312, over 1956387.24 frames. ], batch size: 73, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:50:59,212 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138926.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:51:10,962 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.185e+02 2.764e+02 3.570e+02 5.389e+02, threshold=5.528e+02, percent-clipped=1.0 2022-12-08 14:51:14,883 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0536, 3.8310, 3.5420, 2.8914, 3.4989, 3.7373, 4.1974, 3.4419], device='cuda:1'), covar=tensor([0.0477, 0.0951, 0.0742, 0.1018, 0.0696, 0.0636, 0.0791, 0.0849], device='cuda:1'), in_proj_covar=tensor([0.0152, 0.0166, 0.0138, 0.0124, 0.0142, 0.0153, 0.0136, 0.0140], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:51:35,695 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5746, 3.7093, 3.9292, 3.4585, 3.7598, 3.7216, 1.6338, 3.5734], device='cuda:1'), covar=tensor([0.0427, 0.0400, 0.0311, 0.0492, 0.0340, 0.0425, 0.3086, 0.0310], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0176, 0.0148, 0.0150, 0.0210, 0.0143, 0.0158, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:51:37,497 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138970.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 14:51:40,707 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:51:44,210 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:02,026 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2167, 2.4782, 2.5199, 2.5071, 2.0901, 2.5208, 2.3921, 1.5005], device='cuda:1'), covar=tensor([0.1015, 0.0779, 0.0722, 0.0618, 0.1027, 0.0728, 0.1047, 0.1830], device='cuda:1'), in_proj_covar=tensor([0.0136, 0.0091, 0.0070, 0.0077, 0.0100, 0.0092, 0.0101, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:52:18,361 INFO [train.py:873] (1/4) Epoch 19, batch 2900, loss[loss=0.1043, simple_loss=0.1446, pruned_loss=0.03197, over 14025.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.1401, pruned_loss=0.03349, over 1947620.75 frames. ], batch size: 29, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:52:26,188 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:31,650 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1131, 2.0097, 2.0844, 2.1650, 2.0925, 2.0624, 2.2155, 1.8883], device='cuda:1'), covar=tensor([0.1230, 0.1418, 0.0812, 0.0912, 0.1189, 0.0735, 0.0935, 0.0787], device='cuda:1'), in_proj_covar=tensor([0.0183, 0.0279, 0.0204, 0.0202, 0.0188, 0.0163, 0.0295, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:52:31,854 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-08 14:52:34,841 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6078, 5.1707, 5.0996, 5.5825, 5.1580, 4.6637, 5.5989, 4.6088], device='cuda:1'), covar=tensor([0.0324, 0.0893, 0.0370, 0.0429, 0.0856, 0.0492, 0.0479, 0.0526], device='cuda:1'), in_proj_covar=tensor([0.0183, 0.0280, 0.0204, 0.0202, 0.0188, 0.0163, 0.0295, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 14:52:35,628 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139037.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:38,099 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 1.948e+02 2.422e+02 2.806e+02 5.439e+02, threshold=4.845e+02, percent-clipped=0.0 2022-12-08 14:52:39,984 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139042.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:40,278 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2022-12-08 14:52:49,536 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139053.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:55,732 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.0429, 3.4994, 2.6349, 4.2267, 4.0075, 4.0468, 3.5437, 2.9097], device='cuda:1'), covar=tensor([0.0757, 0.1291, 0.3290, 0.0567, 0.0958, 0.1000, 0.1230, 0.2715], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0288, 0.0259, 0.0292, 0.0323, 0.0302, 0.0256, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:53:02,144 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 14:53:45,725 INFO [train.py:873] (1/4) Epoch 19, batch 3000, loss[loss=0.09766, simple_loss=0.1418, pruned_loss=0.02678, over 14269.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1406, pruned_loss=0.0337, over 1932358.44 frames. ], batch size: 60, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:53:45,726 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 14:53:54,217 INFO [train.py:905] (1/4) Epoch 19, validation: loss=0.142, simple_loss=0.1782, pruned_loss=0.05288, over 857387.00 frames. 2022-12-08 14:53:54,218 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 14:54:14,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.026e+02 2.406e+02 3.345e+02 1.533e+03, threshold=4.812e+02, percent-clipped=5.0 2022-12-08 14:55:22,155 INFO [train.py:873] (1/4) Epoch 19, batch 3100, loss[loss=0.09684, simple_loss=0.1449, pruned_loss=0.02438, over 14302.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.1409, pruned_loss=0.03381, over 1978907.48 frames. ], batch size: 60, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:55:40,046 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:55:41,531 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.036e+02 2.564e+02 2.982e+02 6.918e+02, threshold=5.128e+02, percent-clipped=2.0 2022-12-08 14:56:03,652 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139265.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 14:56:13,049 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139276.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:24,481 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:33,092 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:35,128 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2022-12-08 14:56:48,454 INFO [train.py:873] (1/4) Epoch 19, batch 3200, loss[loss=0.1129, simple_loss=0.1504, pruned_loss=0.03767, over 14428.00 frames. ], tot_loss[loss=0.1031, simple_loss=0.1404, pruned_loss=0.03294, over 2013716.13 frames. ], batch size: 73, lr: 4.14e-03, grad_scale: 8.0 2022-12-08 14:57:04,657 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6183, 2.9271, 2.8574, 2.9756, 2.3478, 3.0093, 2.8279, 1.7365], device='cuda:1'), covar=tensor([0.1138, 0.0739, 0.1053, 0.0613, 0.1015, 0.0565, 0.0916, 0.1798], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0092, 0.0072, 0.0077, 0.0101, 0.0093, 0.0102, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 14:57:06,636 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:06,694 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139337.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 14:57:08,976 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.774e+01 1.930e+02 2.439e+02 3.004e+02 8.294e+02, threshold=4.879e+02, percent-clipped=2.0 2022-12-08 14:57:10,928 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:14,684 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-08 14:57:17,675 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:20,045 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:38,742 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4770, 3.2024, 2.4890, 3.6027, 3.4352, 3.4354, 3.0422, 2.6205], device='cuda:1'), covar=tensor([0.0812, 0.1282, 0.3107, 0.0593, 0.0906, 0.1079, 0.1242, 0.2698], device='cuda:1'), in_proj_covar=tensor([0.0280, 0.0287, 0.0256, 0.0291, 0.0322, 0.0302, 0.0254, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:57:47,842 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:52,092 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:54,115 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 14:58:01,925 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139401.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:58:12,878 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1176, 3.8989, 3.5865, 3.7954, 3.9852, 4.0663, 4.0950, 4.1162], device='cuda:1'), covar=tensor([0.0814, 0.0545, 0.1967, 0.2390, 0.0736, 0.0856, 0.0862, 0.0727], device='cuda:1'), in_proj_covar=tensor([0.0397, 0.0272, 0.0447, 0.0568, 0.0354, 0.0458, 0.0387, 0.0400], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:58:15,482 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7473, 4.8220, 5.2317, 4.5138, 4.9671, 5.2156, 2.0817, 4.5850], device='cuda:1'), covar=tensor([0.0269, 0.0275, 0.0265, 0.0307, 0.0285, 0.0145, 0.2803, 0.0293], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0178, 0.0149, 0.0151, 0.0212, 0.0144, 0.0159, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 14:58:16,250 INFO [train.py:873] (1/4) Epoch 19, batch 3300, loss[loss=0.08872, simple_loss=0.1327, pruned_loss=0.02236, over 14635.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1399, pruned_loss=0.03271, over 2005508.64 frames. ], batch size: 23, lr: 4.14e-03, grad_scale: 8.0 2022-12-08 14:58:35,036 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.023e+02 2.414e+02 3.086e+02 5.603e+02, threshold=4.828e+02, percent-clipped=3.0 2022-12-08 14:58:35,621 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 14:59:41,884 INFO [train.py:873] (1/4) Epoch 19, batch 3400, loss[loss=0.1282, simple_loss=0.161, pruned_loss=0.04771, over 14247.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1396, pruned_loss=0.03304, over 1992348.65 frames. ], batch size: 80, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 14:59:45,855 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1933, 1.4492, 3.2560, 1.6123, 3.0774, 3.2939, 2.2958, 3.4912], device='cuda:1'), covar=tensor([0.0262, 0.2985, 0.0395, 0.2143, 0.1060, 0.0406, 0.1075, 0.0224], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0154, 0.0160, 0.0168, 0.0168, 0.0178, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 14:59:57,873 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5844, 4.6739, 5.0266, 4.3302, 4.7357, 4.8422, 1.8507, 4.4587], device='cuda:1'), covar=tensor([0.0301, 0.0301, 0.0235, 0.0437, 0.0289, 0.0270, 0.2941, 0.0279], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0178, 0.0148, 0.0151, 0.0212, 0.0144, 0.0158, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:00:02,109 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.047e+02 2.601e+02 3.313e+02 6.688e+02, threshold=5.202e+02, percent-clipped=6.0 2022-12-08 15:00:12,636 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8544, 2.2076, 4.7816, 3.1192, 4.6172, 1.9722, 3.2445, 4.6099], device='cuda:1'), covar=tensor([0.0484, 0.4018, 0.0345, 0.6156, 0.0457, 0.3539, 0.1638, 0.0320], device='cuda:1'), in_proj_covar=tensor([0.0251, 0.0197, 0.0219, 0.0268, 0.0240, 0.0201, 0.0200, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 15:00:24,194 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139565.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:00:29,782 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139571.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:00:37,009 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=7.50 vs. limit=5.0 2022-12-08 15:00:39,091 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139582.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:00:49,342 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.5883, 5.4612, 5.0014, 5.6409, 5.1942, 4.9745, 5.6750, 5.4002], device='cuda:1'), covar=tensor([0.0566, 0.0690, 0.0933, 0.0450, 0.0725, 0.0392, 0.0585, 0.0639], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0148, 0.0149, 0.0165, 0.0151, 0.0125, 0.0172, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:00:49,375 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:06,701 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139613.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:10,079 INFO [train.py:873] (1/4) Epoch 19, batch 3500, loss[loss=0.09348, simple_loss=0.1361, pruned_loss=0.02543, over 14221.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1396, pruned_loss=0.03295, over 2020470.41 frames. ], batch size: 60, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:01:13,235 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 15:01:23,117 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139632.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 15:01:23,204 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139632.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:30,130 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.003e+02 2.440e+02 2.964e+02 5.520e+02, threshold=4.880e+02, percent-clipped=2.0 2022-12-08 15:01:32,916 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:34,860 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139645.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:47,365 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6420, 1.6071, 1.5799, 1.6691, 1.7159, 1.0889, 1.4711, 1.4919], device='cuda:1'), covar=tensor([0.0764, 0.0731, 0.0781, 0.0718, 0.0640, 0.1049, 0.1077, 0.0761], device='cuda:1'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0051, 0.0038, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 15:02:25,125 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8357, 3.6165, 3.3214, 2.6193, 3.2759, 3.5194, 3.9555, 3.1847], device='cuda:1'), covar=tensor([0.0501, 0.0853, 0.0761, 0.1025, 0.0724, 0.0566, 0.0581, 0.0843], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0169, 0.0140, 0.0126, 0.0144, 0.0156, 0.0139, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:02:37,619 INFO [train.py:873] (1/4) Epoch 19, batch 3600, loss[loss=0.1078, simple_loss=0.1373, pruned_loss=0.03916, over 10311.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1405, pruned_loss=0.03374, over 1983011.97 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:02:57,975 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.754e+01 2.074e+02 2.504e+02 3.273e+02 8.629e+02, threshold=5.008e+02, percent-clipped=8.0 2022-12-08 15:03:07,804 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7548, 1.9501, 2.1457, 1.9468, 1.8949, 1.7863, 1.8413, 1.2566], device='cuda:1'), covar=tensor([0.0334, 0.0386, 0.0212, 0.0291, 0.0223, 0.0294, 0.0233, 0.0513], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:03:24,067 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139769.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:41,863 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139789.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:55,261 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139804.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:57,302 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:07,191 INFO [train.py:873] (1/4) Epoch 19, batch 3700, loss[loss=0.1235, simple_loss=0.1447, pruned_loss=0.05111, over 3903.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1406, pruned_loss=0.03392, over 1961783.61 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:04:11,851 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-08 15:04:18,388 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:26,980 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 1.959e+02 2.384e+02 2.847e+02 4.025e+02, threshold=4.768e+02, percent-clipped=0.0 2022-12-08 15:04:36,830 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:36,853 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:49,159 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:51,021 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139867.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:06,860 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 15:05:13,008 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8849, 2.8322, 2.1757, 2.9729, 2.8166, 2.8421, 2.5250, 2.2201], device='cuda:1'), covar=tensor([0.0953, 0.1356, 0.2739, 0.0864, 0.1032, 0.1108, 0.1502, 0.2539], device='cuda:1'), in_proj_covar=tensor([0.0280, 0.0287, 0.0256, 0.0291, 0.0321, 0.0302, 0.0254, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:05:14,745 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139894.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:18,875 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8948, 1.5657, 3.0039, 2.6738, 2.8647, 2.9862, 2.2508, 2.9945], device='cuda:1'), covar=tensor([0.1257, 0.1463, 0.0186, 0.0458, 0.0382, 0.0209, 0.0592, 0.0213], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0155, 0.0131, 0.0169, 0.0147, 0.0142, 0.0125, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 15:05:29,288 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:34,120 INFO [train.py:873] (1/4) Epoch 19, batch 3800, loss[loss=0.09641, simple_loss=0.113, pruned_loss=0.03989, over 2639.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1402, pruned_loss=0.03354, over 1962772.73 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:05:43,468 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139927.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:48,157 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139932.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 15:05:53,173 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139938.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:54,736 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.144e+02 2.015e+02 2.567e+02 3.063e+02 4.403e+02, threshold=5.135e+02, percent-clipped=0.0 2022-12-08 15:05:56,575 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139942.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:59,120 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:06:29,964 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:06:41,153 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139993.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:07:06,100 INFO [train.py:873] (1/4) Epoch 19, batch 3900, loss[loss=0.1408, simple_loss=0.151, pruned_loss=0.06528, over 3899.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1403, pruned_loss=0.03359, over 1956593.38 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:07:07,418 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2022-12-08 15:07:10,143 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 15:07:25,634 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.144e+02 2.574e+02 3.333e+02 6.487e+02, threshold=5.147e+02, percent-clipped=6.0 2022-12-08 15:07:52,571 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5137, 2.8077, 2.7477, 2.9146, 2.1501, 2.9161, 2.7119, 1.5115], device='cuda:1'), covar=tensor([0.0878, 0.0756, 0.0957, 0.0462, 0.1050, 0.0592, 0.0914, 0.1844], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0091, 0.0071, 0.0077, 0.0101, 0.0093, 0.0102, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:08:21,797 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140103.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:28,002 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:33,683 INFO [train.py:873] (1/4) Epoch 19, batch 4000, loss[loss=0.1082, simple_loss=0.1489, pruned_loss=0.03372, over 14261.00 frames. ], tot_loss[loss=0.1031, simple_loss=0.1403, pruned_loss=0.03298, over 2027939.70 frames. ], batch size: 39, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:08:41,211 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140125.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:47,806 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.2609, 1.2614, 1.3688, 1.1483, 1.1511, 0.9036, 1.0568, 1.1092], device='cuda:1'), covar=tensor([0.0241, 0.0271, 0.0214, 0.0275, 0.0249, 0.0460, 0.0316, 0.0419], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:08:54,674 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.130e+02 2.579e+02 3.097e+02 6.234e+02, threshold=5.158e+02, percent-clipped=1.0 2022-12-08 15:08:59,204 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:01,028 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7322, 2.3998, 4.7455, 3.1907, 4.5175, 2.3272, 3.4601, 4.5528], device='cuda:1'), covar=tensor([0.0457, 0.3588, 0.0365, 0.5019, 0.0607, 0.3015, 0.1286, 0.0443], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0199, 0.0219, 0.0267, 0.0241, 0.0202, 0.0200, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 15:09:12,071 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140160.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:14,038 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140162.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:15,863 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:22,488 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:52,872 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:02,906 INFO [train.py:873] (1/4) Epoch 19, batch 4100, loss[loss=0.1354, simple_loss=0.1644, pruned_loss=0.05323, over 7770.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1399, pruned_loss=0.03275, over 2037836.58 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:10:11,562 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140227.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:20,975 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:22,567 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.026e+02 2.683e+02 3.306e+02 6.239e+02, threshold=5.366e+02, percent-clipped=2.0 2022-12-08 15:10:39,348 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:52,807 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140275.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:11:00,042 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6393, 2.4756, 2.2701, 2.4049, 2.5656, 2.5856, 2.5849, 2.6108], device='cuda:1'), covar=tensor([0.1252, 0.0864, 0.2796, 0.2705, 0.1201, 0.1291, 0.1544, 0.1040], device='cuda:1'), in_proj_covar=tensor([0.0398, 0.0275, 0.0451, 0.0569, 0.0357, 0.0464, 0.0391, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:11:02,835 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:11:24,650 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9213, 2.1515, 2.9022, 3.0323, 2.8739, 2.2059, 2.9195, 2.4505], device='cuda:1'), covar=tensor([0.0472, 0.1251, 0.0786, 0.0495, 0.0603, 0.1573, 0.0545, 0.0973], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0262, 0.0378, 0.0333, 0.0273, 0.0310, 0.0314, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:11:29,478 INFO [train.py:873] (1/4) Epoch 19, batch 4200, loss[loss=0.09275, simple_loss=0.1364, pruned_loss=0.02457, over 14281.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.14, pruned_loss=0.03291, over 1983119.83 frames. ], batch size: 25, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:11:31,981 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140320.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:11:43,855 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8892, 4.6102, 4.4239, 4.9463, 4.5651, 4.3477, 4.9190, 4.1455], device='cuda:1'), covar=tensor([0.0359, 0.0835, 0.0441, 0.0390, 0.0729, 0.0593, 0.0508, 0.0495], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0276, 0.0204, 0.0201, 0.0187, 0.0161, 0.0293, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:11:49,278 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.052e+02 2.429e+02 3.115e+02 5.934e+02, threshold=4.859e+02, percent-clipped=1.0 2022-12-08 15:11:56,062 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9184, 1.5859, 3.0541, 1.6475, 3.1757, 3.0461, 2.2975, 3.2586], device='cuda:1'), covar=tensor([0.0292, 0.2636, 0.0418, 0.1922, 0.0331, 0.0464, 0.0967, 0.0255], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0155, 0.0161, 0.0169, 0.0167, 0.0179, 0.0133, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:12:04,400 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8947, 3.6668, 3.5690, 3.9409, 3.6391, 3.6298, 3.9553, 3.3004], device='cuda:1'), covar=tensor([0.0532, 0.0975, 0.0531, 0.0461, 0.0810, 0.0986, 0.0587, 0.0583], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0276, 0.0204, 0.0201, 0.0187, 0.0161, 0.0293, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:12:29,888 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140387.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:12:45,791 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6152, 2.3301, 2.9465, 1.8305, 1.9999, 2.6089, 1.5255, 2.6087], device='cuda:1'), covar=tensor([0.1010, 0.1547, 0.0718, 0.2124, 0.2120, 0.0987, 0.3305, 0.1052], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0104, 0.0097, 0.0101, 0.0117, 0.0093, 0.0118, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 15:12:56,648 INFO [train.py:873] (1/4) Epoch 19, batch 4300, loss[loss=0.1396, simple_loss=0.1344, pruned_loss=0.07238, over 1281.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1404, pruned_loss=0.0332, over 1970878.57 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:13:03,512 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:07,832 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1870, 3.9440, 3.8053, 4.2502, 3.9316, 3.7115, 4.2582, 3.6372], device='cuda:1'), covar=tensor([0.0495, 0.0964, 0.0495, 0.0468, 0.0821, 0.1351, 0.0556, 0.0523], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0277, 0.0205, 0.0202, 0.0188, 0.0161, 0.0293, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:13:16,351 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.481e+01 2.038e+02 2.405e+02 2.912e+02 6.757e+02, threshold=4.809e+02, percent-clipped=4.0 2022-12-08 15:13:21,067 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:23,473 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140448.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:33,469 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140459.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:34,396 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140460.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:36,138 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:39,474 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:45,306 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140473.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:59,427 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:02,814 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:14,413 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:16,048 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140508.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:17,022 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140509.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:17,799 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:23,506 INFO [train.py:873] (1/4) Epoch 19, batch 4400, loss[loss=0.1515, simple_loss=0.1411, pruned_loss=0.081, over 1326.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1412, pruned_loss=0.03414, over 1982473.84 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:14:41,664 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.0711, 2.8756, 2.6651, 2.8252, 3.0261, 3.0253, 3.0569, 3.0716], device='cuda:1'), covar=tensor([0.1142, 0.0894, 0.2558, 0.2586, 0.1000, 0.1159, 0.1301, 0.0926], device='cuda:1'), in_proj_covar=tensor([0.0395, 0.0272, 0.0448, 0.0566, 0.0355, 0.0459, 0.0390, 0.0399], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:14:43,958 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.073e+02 2.032e+02 2.425e+02 2.812e+02 5.381e+02, threshold=4.850e+02, percent-clipped=2.0 2022-12-08 15:14:52,670 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140550.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:56,025 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140554.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:09,638 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140570.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:28,362 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8823, 1.5377, 3.2400, 2.8607, 3.0834, 3.2456, 2.4999, 3.2360], device='cuda:1'), covar=tensor([0.1473, 0.1652, 0.0182, 0.0430, 0.0352, 0.0215, 0.0422, 0.0184], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0159, 0.0133, 0.0172, 0.0149, 0.0144, 0.0128, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 15:15:50,190 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140615.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:51,859 INFO [train.py:873] (1/4) Epoch 19, batch 4500, loss[loss=0.1255, simple_loss=0.1423, pruned_loss=0.05434, over 3854.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1405, pruned_loss=0.03346, over 1974416.09 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 4.0 2022-12-08 15:16:12,049 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 2.100e+02 2.600e+02 3.124e+02 5.170e+02, threshold=5.201e+02, percent-clipped=3.0 2022-12-08 15:16:15,010 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4810, 4.2105, 3.9956, 4.5011, 4.1813, 3.9520, 4.5155, 3.7452], device='cuda:1'), covar=tensor([0.0410, 0.0864, 0.0494, 0.0448, 0.0768, 0.0912, 0.0561, 0.0576], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0277, 0.0205, 0.0202, 0.0187, 0.0161, 0.0293, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:16:29,900 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3493, 3.1517, 2.9202, 3.0637, 3.2709, 3.2965, 3.3104, 3.3319], device='cuda:1'), covar=tensor([0.0920, 0.0690, 0.2180, 0.2494, 0.0887, 0.0965, 0.1135, 0.0867], device='cuda:1'), in_proj_covar=tensor([0.0397, 0.0274, 0.0449, 0.0567, 0.0355, 0.0461, 0.0393, 0.0401], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:16:44,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7828, 2.9208, 2.9925, 2.9057, 2.9181, 2.6771, 1.6526, 2.7111], device='cuda:1'), covar=tensor([0.0493, 0.0409, 0.0350, 0.0396, 0.0354, 0.0942, 0.2554, 0.0341], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0179, 0.0150, 0.0152, 0.0212, 0.0146, 0.0160, 0.0200], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:17:04,727 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 15:17:17,738 INFO [train.py:873] (1/4) Epoch 19, batch 4600, loss[loss=0.121, simple_loss=0.1399, pruned_loss=0.05106, over 3898.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1406, pruned_loss=0.0339, over 1929088.09 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 4.0 2022-12-08 15:17:23,436 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140723.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:17:25,391 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=5.97 vs. limit=5.0 2022-12-08 15:17:36,410 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4491, 3.5672, 3.7755, 3.4494, 3.6226, 3.5573, 1.5703, 3.4522], device='cuda:1'), covar=tensor([0.0414, 0.0419, 0.0320, 0.0448, 0.0308, 0.0490, 0.2990, 0.0305], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0179, 0.0149, 0.0152, 0.0211, 0.0145, 0.0160, 0.0199], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:17:39,454 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.073e+02 2.575e+02 3.535e+02 1.946e+03, threshold=5.149e+02, percent-clipped=6.0 2022-12-08 15:17:41,307 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140743.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:17:55,001 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140759.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:17:58,704 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 15:18:00,987 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:17,137 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:33,901 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 15:18:36,904 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140807.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:38,956 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140809.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:43,014 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:45,758 INFO [train.py:873] (1/4) Epoch 19, batch 4700, loss[loss=0.08282, simple_loss=0.13, pruned_loss=0.01782, over 14311.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1406, pruned_loss=0.03345, over 1983228.44 frames. ], batch size: 31, lr: 4.11e-03, grad_scale: 4.0 2022-12-08 15:19:00,741 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8829, 1.6021, 1.9653, 1.6053, 1.9565, 1.8149, 1.6382, 1.8258], device='cuda:1'), covar=tensor([0.0672, 0.1421, 0.0511, 0.0553, 0.0577, 0.0855, 0.0404, 0.0424], device='cuda:1'), in_proj_covar=tensor([0.0347, 0.0306, 0.0384, 0.0296, 0.0361, 0.0320, 0.0360, 0.0294], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:19:06,570 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.144e+02 2.562e+02 3.057e+02 6.222e+02, threshold=5.124e+02, percent-clipped=4.0 2022-12-08 15:19:10,081 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140845.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:19:17,596 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.52 vs. limit=5.0 2022-12-08 15:19:24,166 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1712, 1.9478, 2.2517, 2.3351, 1.9058, 1.9962, 2.1949, 2.1408], device='cuda:1'), covar=tensor([0.0376, 0.0736, 0.0376, 0.0372, 0.0624, 0.0869, 0.0475, 0.0457], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0264, 0.0381, 0.0336, 0.0275, 0.0311, 0.0316, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:19:28,317 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:19:31,743 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2232, 3.7846, 2.8539, 4.3426, 4.0666, 4.2070, 3.7306, 3.0998], device='cuda:1'), covar=tensor([0.0569, 0.1064, 0.2943, 0.0480, 0.0878, 0.0995, 0.1003, 0.2505], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0290, 0.0260, 0.0292, 0.0323, 0.0302, 0.0256, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:19:32,426 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:19:47,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5167, 4.0416, 3.0873, 4.7402, 4.2204, 4.5304, 3.9205, 3.3340], device='cuda:1'), covar=tensor([0.0575, 0.0925, 0.2889, 0.0388, 0.0801, 0.0975, 0.1076, 0.2460], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0290, 0.0260, 0.0293, 0.0324, 0.0303, 0.0257, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:20:11,792 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140915.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:13,317 INFO [train.py:873] (1/4) Epoch 19, batch 4800, loss[loss=0.1081, simple_loss=0.1451, pruned_loss=0.03558, over 14256.00 frames. ], tot_loss[loss=0.103, simple_loss=0.1401, pruned_loss=0.03295, over 1938426.85 frames. ], batch size: 76, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:20:13,515 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140917.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:18,728 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7042, 3.8781, 4.0590, 3.7278, 3.8747, 3.9006, 1.6703, 3.7112], device='cuda:1'), covar=tensor([0.0347, 0.0351, 0.0302, 0.0402, 0.0319, 0.0353, 0.3017, 0.0302], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0176, 0.0148, 0.0150, 0.0209, 0.0144, 0.0158, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:20:24,275 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9945, 2.0691, 2.0821, 2.0643, 2.0221, 1.6329, 1.2912, 1.8435], device='cuda:1'), covar=tensor([0.0749, 0.0571, 0.0510, 0.0424, 0.0532, 0.1607, 0.2452, 0.0534], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0176, 0.0148, 0.0150, 0.0209, 0.0144, 0.0158, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:20:34,741 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.115e+02 2.464e+02 2.978e+02 7.209e+02, threshold=4.928e+02, percent-clipped=2.0 2022-12-08 15:20:39,972 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140947.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:53,894 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140963.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:21:07,419 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:21:07,657 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 15:21:34,093 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:21:41,896 INFO [train.py:873] (1/4) Epoch 19, batch 4900, loss[loss=0.09864, simple_loss=0.1451, pruned_loss=0.02608, over 14355.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1405, pruned_loss=0.03313, over 1922255.88 frames. ], batch size: 28, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:22:02,359 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.070e+02 2.764e+02 3.531e+02 1.295e+03, threshold=5.527e+02, percent-clipped=7.0 2022-12-08 15:22:04,091 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:22:34,920 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141079.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:22:45,501 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141091.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:08,040 INFO [train.py:873] (1/4) Epoch 19, batch 5000, loss[loss=0.12, simple_loss=0.1323, pruned_loss=0.05382, over 2674.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.14, pruned_loss=0.03252, over 1910673.21 frames. ], batch size: 100, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:23:21,311 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 15:23:28,878 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.058e+02 2.477e+02 3.160e+02 7.208e+02, threshold=4.953e+02, percent-clipped=1.0 2022-12-08 15:23:30,738 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141143.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:32,329 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:49,061 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:49,116 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:13,406 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:22,791 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141204.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:30,574 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:34,008 INFO [train.py:873] (1/4) Epoch 19, batch 5100, loss[loss=0.09517, simple_loss=0.1392, pruned_loss=0.02556, over 14371.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1397, pruned_loss=0.03261, over 1965784.25 frames. ], batch size: 18, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:24:35,917 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9537, 4.5880, 4.5052, 4.9935, 4.6663, 4.3841, 4.9638, 4.2403], device='cuda:1'), covar=tensor([0.0384, 0.0971, 0.0378, 0.0367, 0.0677, 0.0559, 0.0508, 0.0478], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0277, 0.0203, 0.0201, 0.0186, 0.0162, 0.0292, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:24:54,060 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.078e+02 2.511e+02 3.062e+02 5.602e+02, threshold=5.021e+02, percent-clipped=1.0 2022-12-08 15:25:21,719 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:25:47,944 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141303.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:25:59,794 INFO [train.py:873] (1/4) Epoch 19, batch 5200, loss[loss=0.1052, simple_loss=0.1374, pruned_loss=0.03643, over 6036.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1404, pruned_loss=0.03317, over 2002493.82 frames. ], batch size: 100, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:26:10,194 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:26:20,994 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.014e+02 2.549e+02 3.340e+02 7.347e+02, threshold=5.099e+02, percent-clipped=2.0 2022-12-08 15:26:35,938 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2858, 3.0086, 2.3884, 3.4246, 3.2625, 3.2696, 2.8446, 2.3617], device='cuda:1'), covar=tensor([0.0859, 0.1431, 0.3010, 0.0701, 0.0882, 0.1035, 0.1490, 0.3338], device='cuda:1'), in_proj_covar=tensor([0.0279, 0.0287, 0.0256, 0.0290, 0.0320, 0.0301, 0.0254, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:26:54,236 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141379.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:04,106 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:08,653 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 15:27:13,912 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2263, 3.0658, 2.7941, 2.9815, 3.1556, 3.1877, 3.1835, 3.2150], device='cuda:1'), covar=tensor([0.1040, 0.0655, 0.2299, 0.2356, 0.0960, 0.1086, 0.1237, 0.0869], device='cuda:1'), in_proj_covar=tensor([0.0398, 0.0277, 0.0454, 0.0568, 0.0358, 0.0462, 0.0398, 0.0404], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:27:18,948 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9801, 1.5358, 2.0015, 1.4052, 1.7405, 2.0725, 1.9071, 1.7859], device='cuda:1'), covar=tensor([0.1050, 0.0693, 0.0971, 0.1542, 0.1501, 0.0975, 0.0989, 0.1518], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0171, 0.0142, 0.0127, 0.0146, 0.0157, 0.0141, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:27:28,119 INFO [train.py:873] (1/4) Epoch 19, batch 5300, loss[loss=0.07666, simple_loss=0.1245, pruned_loss=0.0144, over 14250.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1397, pruned_loss=0.0327, over 1975157.72 frames. ], batch size: 39, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:27:28,320 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141417.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:27:36,865 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141427.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:38,746 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3793, 2.6504, 2.6126, 2.6996, 2.2312, 2.6589, 2.5016, 1.6273], device='cuda:1'), covar=tensor([0.0939, 0.0688, 0.0752, 0.0667, 0.0920, 0.0682, 0.0952, 0.1787], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0093, 0.0072, 0.0077, 0.0102, 0.0093, 0.0103, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 15:27:48,827 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.160e+02 2.222e+02 2.556e+02 2.994e+02 5.858e+02, threshold=5.112e+02, percent-clipped=4.0 2022-12-08 15:27:48,990 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:10,409 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141465.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:21,408 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141478.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 15:28:39,744 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141499.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:41,048 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2022-12-08 15:28:42,437 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141502.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:51,518 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:54,901 INFO [train.py:873] (1/4) Epoch 19, batch 5400, loss[loss=0.09264, simple_loss=0.1311, pruned_loss=0.02708, over 11157.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.14, pruned_loss=0.03357, over 1892734.29 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:29:01,442 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 15:29:01,915 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3191, 3.4866, 4.2520, 3.0844, 2.4945, 3.6387, 2.0444, 3.6749], device='cuda:1'), covar=tensor([0.1875, 0.0899, 0.0597, 0.2107, 0.1952, 0.0793, 0.2931, 0.1187], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0104, 0.0098, 0.0102, 0.0115, 0.0093, 0.0118, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 15:29:16,144 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.017e+02 2.549e+02 3.465e+02 7.895e+02, threshold=5.098e+02, percent-clipped=4.0 2022-12-08 15:29:19,988 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3778, 1.0073, 1.1953, 0.8670, 1.1202, 1.4054, 1.0455, 1.1179], device='cuda:1'), covar=tensor([0.0578, 0.0961, 0.0863, 0.0620, 0.1064, 0.0849, 0.0695, 0.1381], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0169, 0.0140, 0.0125, 0.0145, 0.0155, 0.0139, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:29:43,888 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141573.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:30:10,137 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:30:22,125 INFO [train.py:873] (1/4) Epoch 19, batch 5500, loss[loss=0.0943, simple_loss=0.1389, pruned_loss=0.02486, over 14393.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.14, pruned_loss=0.03326, over 1928229.65 frames. ], batch size: 44, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:30:25,593 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141621.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:30:43,271 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.230e+02 2.074e+02 2.608e+02 3.444e+02 6.357e+02, threshold=5.216e+02, percent-clipped=6.0 2022-12-08 15:30:46,405 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2022-12-08 15:30:52,127 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141651.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:31:21,361 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:31:48,909 INFO [train.py:873] (1/4) Epoch 19, batch 5600, loss[loss=0.1088, simple_loss=0.117, pruned_loss=0.05027, over 2658.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.1401, pruned_loss=0.03328, over 1898294.76 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:32:09,418 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.697e+01 2.227e+02 2.853e+02 3.557e+02 6.540e+02, threshold=5.705e+02, percent-clipped=5.0 2022-12-08 15:32:27,068 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9406, 4.5433, 4.3773, 4.9348, 4.5955, 4.3816, 4.9119, 4.1799], device='cuda:1'), covar=tensor([0.0352, 0.0931, 0.0435, 0.0384, 0.0741, 0.0555, 0.0529, 0.0457], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0280, 0.0204, 0.0202, 0.0188, 0.0162, 0.0294, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:32:31,286 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0466, 1.6916, 2.0588, 1.4557, 1.8252, 2.1301, 1.9766, 1.8756], device='cuda:1'), covar=tensor([0.0992, 0.0791, 0.0936, 0.1477, 0.1463, 0.0957, 0.0794, 0.1521], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0170, 0.0141, 0.0126, 0.0146, 0.0156, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:32:37,281 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141773.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 15:32:58,707 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:33:00,362 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141799.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:33:16,298 INFO [train.py:873] (1/4) Epoch 19, batch 5700, loss[loss=0.09025, simple_loss=0.1348, pruned_loss=0.02285, over 14370.00 frames. ], tot_loss[loss=0.1032, simple_loss=0.14, pruned_loss=0.03321, over 1905076.97 frames. ], batch size: 31, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:33:36,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.084e+02 2.066e+02 2.688e+02 3.167e+02 5.846e+02, threshold=5.376e+02, percent-clipped=2.0 2022-12-08 15:33:41,764 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141847.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:34:41,422 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:34:42,167 INFO [train.py:873] (1/4) Epoch 19, batch 5800, loss[loss=0.1331, simple_loss=0.1329, pruned_loss=0.06666, over 2618.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1406, pruned_loss=0.0338, over 1917849.21 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:34:51,867 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141928.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:03,273 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.310e+01 2.008e+02 2.595e+02 3.223e+02 5.261e+02, threshold=5.190e+02, percent-clipped=0.0 2022-12-08 15:35:35,498 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141977.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:43,118 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:46,557 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:36:00,700 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2329, 1.8621, 3.3085, 2.4017, 3.2065, 1.8348, 2.5639, 3.2384], device='cuda:1'), covar=tensor([0.0748, 0.3654, 0.0668, 0.3973, 0.0826, 0.2952, 0.1308, 0.0601], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0198, 0.0220, 0.0266, 0.0240, 0.0201, 0.0200, 0.0222], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 15:36:11,183 INFO [train.py:873] (1/4) Epoch 19, batch 5900, loss[loss=0.1127, simple_loss=0.1547, pruned_loss=0.03532, over 14412.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1398, pruned_loss=0.03269, over 1967046.10 frames. ], batch size: 53, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:36:24,799 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:36:27,583 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9311, 2.8350, 2.2553, 2.9885, 2.8510, 2.8637, 2.5927, 2.2786], device='cuda:1'), covar=tensor([0.0820, 0.1119, 0.2484, 0.0743, 0.1175, 0.0945, 0.1274, 0.2237], device='cuda:1'), in_proj_covar=tensor([0.0280, 0.0286, 0.0256, 0.0291, 0.0320, 0.0302, 0.0254, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:36:31,705 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 2.065e+02 2.424e+02 3.179e+02 4.807e+02, threshold=4.849e+02, percent-clipped=0.0 2022-12-08 15:36:52,019 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7305, 2.8826, 4.4510, 3.4511, 4.4406, 4.2627, 4.2512, 3.8846], device='cuda:1'), covar=tensor([0.0602, 0.2833, 0.0845, 0.1492, 0.0636, 0.0950, 0.1400, 0.1505], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0308, 0.0385, 0.0296, 0.0363, 0.0321, 0.0359, 0.0295], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:37:00,281 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142073.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:37:10,397 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1302, 2.4985, 2.2142, 2.3255, 2.2021, 2.0364, 1.8246, 2.1069], device='cuda:1'), covar=tensor([0.0301, 0.0239, 0.0298, 0.0323, 0.0311, 0.0475, 0.0464, 0.0423], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:37:21,271 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:37:21,657 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 15:37:38,674 INFO [train.py:873] (1/4) Epoch 19, batch 6000, loss[loss=0.09311, simple_loss=0.1322, pruned_loss=0.02703, over 14165.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1393, pruned_loss=0.03302, over 1904518.02 frames. ], batch size: 99, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:37:38,674 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 15:37:47,304 INFO [train.py:905] (1/4) Epoch 19, validation: loss=0.1418, simple_loss=0.1782, pruned_loss=0.05266, over 857387.00 frames. 2022-12-08 15:37:47,305 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 15:37:50,881 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142121.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:38:08,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 1.928e+02 2.466e+02 3.087e+02 6.019e+02, threshold=4.931e+02, percent-clipped=3.0 2022-12-08 15:38:10,165 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3240, 0.8659, 1.1439, 0.8000, 1.0083, 1.2813, 1.0068, 1.0449], device='cuda:1'), covar=tensor([0.0574, 0.1122, 0.0929, 0.0653, 0.1478, 0.1344, 0.0776, 0.1856], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0171, 0.0142, 0.0126, 0.0147, 0.0157, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:38:11,906 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:39:11,445 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 15:39:15,264 INFO [train.py:873] (1/4) Epoch 19, batch 6100, loss[loss=0.1013, simple_loss=0.1411, pruned_loss=0.03077, over 14169.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1393, pruned_loss=0.03292, over 1883453.48 frames. ], batch size: 84, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:39:36,596 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.000e+02 2.533e+02 3.049e+02 1.114e+03, threshold=5.065e+02, percent-clipped=3.0 2022-12-08 15:39:53,289 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7022, 3.2612, 3.4421, 3.7082, 3.5703, 3.6305, 3.6650, 3.0989], device='cuda:1'), covar=tensor([0.1126, 0.1969, 0.1133, 0.0977, 0.1188, 0.1282, 0.1243, 0.1211], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0280, 0.0205, 0.0204, 0.0187, 0.0163, 0.0295, 0.0173], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:40:03,700 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:40:14,735 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:40:35,170 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0798, 2.2328, 2.2853, 2.3908, 2.0645, 2.3715, 2.2132, 1.4505], device='cuda:1'), covar=tensor([0.0939, 0.0842, 0.0489, 0.0579, 0.0963, 0.0547, 0.0990, 0.1730], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0092, 0.0072, 0.0077, 0.0101, 0.0093, 0.0103, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:40:42,867 INFO [train.py:873] (1/4) Epoch 19, batch 6200, loss[loss=0.1124, simple_loss=0.1384, pruned_loss=0.04321, over 6947.00 frames. ], tot_loss[loss=0.103, simple_loss=0.14, pruned_loss=0.03298, over 1973533.59 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:40:52,700 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1240, 2.2513, 2.3755, 2.3112, 2.1797, 1.9326, 1.8131, 1.6689], device='cuda:1'), covar=tensor([0.0344, 0.0608, 0.0252, 0.0330, 0.0268, 0.0447, 0.0399, 0.0609], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:41:04,426 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.036e+02 2.651e+02 3.192e+02 8.704e+02, threshold=5.303e+02, percent-clipped=4.0 2022-12-08 15:42:11,862 INFO [train.py:873] (1/4) Epoch 19, batch 6300, loss[loss=0.09869, simple_loss=0.1322, pruned_loss=0.03257, over 6954.00 frames. ], tot_loss[loss=0.1032, simple_loss=0.1399, pruned_loss=0.03321, over 1933396.60 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:42:30,849 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6245, 2.0511, 2.6620, 2.7486, 2.5297, 2.0546, 2.6664, 2.2559], device='cuda:1'), covar=tensor([0.0580, 0.1392, 0.0703, 0.0549, 0.0795, 0.1714, 0.0585, 0.0960], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0261, 0.0377, 0.0330, 0.0272, 0.0308, 0.0313, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:42:32,398 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.018e+02 2.111e+02 2.568e+02 3.235e+02 6.197e+02, threshold=5.136e+02, percent-clipped=2.0 2022-12-08 15:42:44,570 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8225, 1.1327, 1.2921, 1.2686, 0.9186, 1.2991, 1.0680, 0.7709], device='cuda:1'), covar=tensor([0.1533, 0.0887, 0.0450, 0.0476, 0.1803, 0.0870, 0.1333, 0.1419], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0091, 0.0071, 0.0076, 0.0100, 0.0092, 0.0102, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:42:55,693 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5939, 1.5669, 1.5666, 1.4065, 1.3648, 1.3263, 1.1974, 1.1965], device='cuda:1'), covar=tensor([0.0179, 0.0199, 0.0180, 0.0200, 0.0213, 0.0339, 0.0257, 0.0363], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0023, 0.0035, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:43:05,506 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1860, 2.0374, 2.1225, 2.2064, 2.1309, 2.1196, 2.2533, 1.9265], device='cuda:1'), covar=tensor([0.0929, 0.1239, 0.0788, 0.0858, 0.0975, 0.0673, 0.0939, 0.0707], device='cuda:1'), in_proj_covar=tensor([0.0184, 0.0283, 0.0207, 0.0206, 0.0190, 0.0165, 0.0299, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:43:20,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7726, 1.9507, 3.7349, 2.6952, 3.6583, 1.9002, 2.7333, 3.6556], device='cuda:1'), covar=tensor([0.0778, 0.3967, 0.0572, 0.4476, 0.0726, 0.3213, 0.1554, 0.0572], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0198, 0.0219, 0.0268, 0.0240, 0.0201, 0.0201, 0.0222], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 15:43:38,807 INFO [train.py:873] (1/4) Epoch 19, batch 6400, loss[loss=0.1171, simple_loss=0.1476, pruned_loss=0.04328, over 7792.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1391, pruned_loss=0.032, over 1925009.37 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:44:00,396 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 1.982e+02 2.595e+02 3.310e+02 8.116e+02, threshold=5.191e+02, percent-clipped=3.0 2022-12-08 15:44:15,850 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.6128, 4.3974, 4.1336, 4.5677, 4.1474, 3.8896, 4.6716, 4.4199], device='cuda:1'), covar=tensor([0.0504, 0.0786, 0.0802, 0.0513, 0.0776, 0.0590, 0.0473, 0.0606], device='cuda:1'), in_proj_covar=tensor([0.0148, 0.0147, 0.0150, 0.0164, 0.0152, 0.0127, 0.0171, 0.0151], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:44:27,715 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142572.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:44:28,662 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142573.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:44:38,118 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142584.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:07,274 INFO [train.py:873] (1/4) Epoch 19, batch 6500, loss[loss=0.1417, simple_loss=0.1606, pruned_loss=0.06142, over 8634.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1389, pruned_loss=0.03198, over 1972469.05 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 16.0 2022-12-08 15:45:09,993 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142620.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:14,295 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2987, 2.2275, 3.2276, 2.4487, 3.2705, 3.1565, 3.0373, 2.7613], device='cuda:1'), covar=tensor([0.0923, 0.2911, 0.1094, 0.1837, 0.0775, 0.1016, 0.1218, 0.1647], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0308, 0.0385, 0.0297, 0.0363, 0.0321, 0.0359, 0.0295], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:45:19,975 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142632.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:21,819 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142634.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:27,970 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.114e+02 2.216e+02 2.673e+02 3.413e+02 8.329e+02, threshold=5.346e+02, percent-clipped=5.0 2022-12-08 15:45:29,935 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9578, 1.7122, 4.4036, 4.0586, 4.0783, 4.4726, 3.9514, 4.4884], device='cuda:1'), covar=tensor([0.1714, 0.1770, 0.0141, 0.0303, 0.0270, 0.0189, 0.0280, 0.0133], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0157, 0.0132, 0.0169, 0.0149, 0.0143, 0.0127, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 15:45:46,004 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1121, 1.2920, 1.2978, 1.1083, 0.8791, 1.0569, 0.8928, 1.2644], device='cuda:1'), covar=tensor([0.1831, 0.2489, 0.1417, 0.2529, 0.3165, 0.1539, 0.1956, 0.1364], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0104, 0.0098, 0.0102, 0.0116, 0.0093, 0.0117, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 15:45:58,001 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1527, 4.8883, 4.4737, 4.7529, 4.7900, 5.0442, 5.0880, 5.0645], device='cuda:1'), covar=tensor([0.0611, 0.0430, 0.2003, 0.2403, 0.0659, 0.0731, 0.0776, 0.0771], device='cuda:1'), in_proj_covar=tensor([0.0397, 0.0280, 0.0456, 0.0575, 0.0361, 0.0465, 0.0399, 0.0404], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:46:01,849 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-08 15:46:34,423 INFO [train.py:873] (1/4) Epoch 19, batch 6600, loss[loss=0.1104, simple_loss=0.1483, pruned_loss=0.03629, over 14375.00 frames. ], tot_loss[loss=0.1017, simple_loss=0.139, pruned_loss=0.03218, over 1990714.94 frames. ], batch size: 73, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:46:35,194 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2022-12-08 15:46:40,660 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142724.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:46:42,854 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7811, 1.1055, 1.2657, 1.2190, 0.9607, 1.2709, 1.1268, 0.8101], device='cuda:1'), covar=tensor([0.1697, 0.1132, 0.0626, 0.0596, 0.1989, 0.1184, 0.1604, 0.1637], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0091, 0.0071, 0.0076, 0.0101, 0.0092, 0.0102, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:46:49,537 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-08 15:46:56,546 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.141e+02 2.630e+02 3.221e+02 9.486e+02, threshold=5.260e+02, percent-clipped=4.0 2022-12-08 15:47:30,802 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3878, 2.4198, 2.4326, 2.1715, 2.1017, 1.9896, 1.7873, 1.8783], device='cuda:1'), covar=tensor([0.0263, 0.0376, 0.0236, 0.0571, 0.0460, 0.0530, 0.0544, 0.0607], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 15:47:33,950 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 15:47:34,379 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142785.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:48:02,762 INFO [train.py:873] (1/4) Epoch 19, batch 6700, loss[loss=0.1253, simple_loss=0.1508, pruned_loss=0.04987, over 7770.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1391, pruned_loss=0.03227, over 1947432.39 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:48:23,750 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 1.995e+02 2.478e+02 2.934e+02 5.263e+02, threshold=4.957e+02, percent-clipped=1.0 2022-12-08 15:49:19,885 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4379, 2.7327, 4.3341, 4.6203, 4.2992, 2.5078, 4.5213, 3.5795], device='cuda:1'), covar=tensor([0.0464, 0.1297, 0.1010, 0.0419, 0.0544, 0.2327, 0.0429, 0.1021], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0262, 0.0377, 0.0331, 0.0272, 0.0308, 0.0313, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:49:28,806 INFO [train.py:873] (1/4) Epoch 19, batch 6800, loss[loss=0.1215, simple_loss=0.1453, pruned_loss=0.0489, over 4942.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1393, pruned_loss=0.03215, over 1992398.58 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:49:39,758 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:49:45,970 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8467, 2.8536, 4.6356, 3.4674, 4.5638, 4.2542, 4.3074, 3.9085], device='cuda:1'), covar=tensor([0.0594, 0.3071, 0.0799, 0.1556, 0.0702, 0.1071, 0.1399, 0.1401], device='cuda:1'), in_proj_covar=tensor([0.0349, 0.0309, 0.0386, 0.0298, 0.0363, 0.0322, 0.0362, 0.0296], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:49:50,756 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.071e+02 1.868e+02 2.461e+02 3.045e+02 8.260e+02, threshold=4.922e+02, percent-clipped=6.0 2022-12-08 15:50:03,274 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 15:50:24,077 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5640, 1.9415, 4.4273, 2.4719, 4.3127, 4.6590, 4.2027, 5.0462], device='cuda:1'), covar=tensor([0.0233, 0.2909, 0.0415, 0.1835, 0.0325, 0.0395, 0.0365, 0.0173], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0154, 0.0159, 0.0166, 0.0165, 0.0177, 0.0132, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:50:35,557 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2022-12-08 15:50:53,942 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 15:50:56,727 INFO [train.py:873] (1/4) Epoch 19, batch 6900, loss[loss=0.1024, simple_loss=0.1306, pruned_loss=0.03711, over 7768.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1394, pruned_loss=0.0323, over 2015241.86 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:51:18,509 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 2.146e+02 2.489e+02 3.023e+02 7.661e+02, threshold=4.978e+02, percent-clipped=3.0 2022-12-08 15:51:51,285 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:52:23,546 INFO [train.py:873] (1/4) Epoch 19, batch 7000, loss[loss=0.1176, simple_loss=0.1218, pruned_loss=0.05675, over 1173.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1396, pruned_loss=0.03271, over 2004440.61 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:52:23,735 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1542, 2.0465, 2.2465, 2.3905, 2.0401, 1.9806, 2.2175, 2.1635], device='cuda:1'), covar=tensor([0.0375, 0.0752, 0.0339, 0.0324, 0.0537, 0.0787, 0.0443, 0.0404], device='cuda:1'), in_proj_covar=tensor([0.0296, 0.0263, 0.0378, 0.0332, 0.0273, 0.0310, 0.0315, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:52:35,415 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9812, 3.7746, 3.4336, 2.8429, 3.4082, 3.6900, 4.1157, 3.3403], device='cuda:1'), covar=tensor([0.0513, 0.0801, 0.0795, 0.1056, 0.0802, 0.0534, 0.0760, 0.0885], device='cuda:1'), in_proj_covar=tensor([0.0154, 0.0168, 0.0140, 0.0125, 0.0146, 0.0155, 0.0139, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:52:46,400 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 1.976e+02 2.529e+02 3.068e+02 1.002e+03, threshold=5.057e+02, percent-clipped=4.0 2022-12-08 15:52:48,894 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5902, 1.8142, 4.4148, 2.0493, 4.3377, 4.6408, 4.2139, 5.0717], device='cuda:1'), covar=tensor([0.0267, 0.3114, 0.0396, 0.2243, 0.0335, 0.0442, 0.0356, 0.0170], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0155, 0.0161, 0.0168, 0.0167, 0.0179, 0.0134, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:53:31,902 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3056, 3.1722, 3.9677, 2.9276, 2.5044, 3.3301, 2.1594, 3.3244], device='cuda:1'), covar=tensor([0.1163, 0.0920, 0.0500, 0.1454, 0.1805, 0.0835, 0.2531, 0.1223], device='cuda:1'), in_proj_covar=tensor([0.0087, 0.0103, 0.0096, 0.0101, 0.0115, 0.0092, 0.0116, 0.0096], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 15:53:50,607 INFO [train.py:873] (1/4) Epoch 19, batch 7100, loss[loss=0.1112, simple_loss=0.1467, pruned_loss=0.03784, over 14469.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1391, pruned_loss=0.03245, over 1912703.87 frames. ], batch size: 51, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:54:00,948 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:54:10,222 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3906, 4.5021, 4.8057, 4.0697, 4.5758, 4.7923, 1.9084, 4.3359], device='cuda:1'), covar=tensor([0.0369, 0.0349, 0.0306, 0.0557, 0.0314, 0.0221, 0.2970, 0.0289], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0175, 0.0147, 0.0149, 0.0208, 0.0143, 0.0157, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:54:12,484 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.837e+01 2.042e+02 2.415e+02 3.007e+02 1.079e+03, threshold=4.829e+02, percent-clipped=5.0 2022-12-08 15:54:36,168 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.8789, 2.6948, 2.7608, 2.8924, 2.7686, 2.8187, 2.9573, 2.5368], device='cuda:1'), covar=tensor([0.0833, 0.1318, 0.0685, 0.0732, 0.0949, 0.0644, 0.0835, 0.0767], device='cuda:1'), in_proj_covar=tensor([0.0185, 0.0283, 0.0208, 0.0204, 0.0190, 0.0164, 0.0298, 0.0174], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 15:54:42,673 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=143277.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:54:55,837 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3799, 2.6924, 4.2617, 4.4564, 4.2420, 2.5606, 4.5406, 3.4359], device='cuda:1'), covar=tensor([0.0447, 0.1336, 0.1021, 0.0439, 0.0512, 0.2077, 0.0433, 0.0976], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0261, 0.0376, 0.0330, 0.0271, 0.0308, 0.0312, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 15:55:17,811 INFO [train.py:873] (1/4) Epoch 19, batch 7200, loss[loss=0.1252, simple_loss=0.124, pruned_loss=0.06315, over 1178.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1394, pruned_loss=0.03236, over 1928657.38 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:55:22,376 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5149, 3.1290, 3.0244, 2.0795, 2.9117, 3.1842, 3.4405, 2.7780], device='cuda:1'), covar=tensor([0.0640, 0.0855, 0.0865, 0.1331, 0.0890, 0.0625, 0.0702, 0.1112], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0170, 0.0141, 0.0126, 0.0146, 0.0156, 0.0140, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 15:55:37,577 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2171, 1.3883, 4.1622, 2.0463, 4.0397, 4.2599, 3.5209, 4.5842], device='cuda:1'), covar=tensor([0.0254, 0.3415, 0.0408, 0.2161, 0.0410, 0.0398, 0.0614, 0.0177], device='cuda:1'), in_proj_covar=tensor([0.0172, 0.0155, 0.0160, 0.0167, 0.0165, 0.0177, 0.0133, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:55:40,829 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.274e+02 2.664e+02 3.314e+02 6.390e+02, threshold=5.328e+02, percent-clipped=7.0 2022-12-08 15:56:13,727 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:56:18,953 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.43 vs. limit=2.0 2022-12-08 15:56:30,705 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7579, 4.5849, 4.1612, 4.3903, 4.5256, 4.6688, 4.7358, 4.7295], device='cuda:1'), covar=tensor([0.0854, 0.0499, 0.2322, 0.2768, 0.0834, 0.0826, 0.0923, 0.0811], device='cuda:1'), in_proj_covar=tensor([0.0399, 0.0281, 0.0460, 0.0579, 0.0365, 0.0467, 0.0401, 0.0405], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:56:30,789 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143400.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:56:46,181 INFO [train.py:873] (1/4) Epoch 19, batch 7300, loss[loss=0.09583, simple_loss=0.1395, pruned_loss=0.02606, over 14258.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1388, pruned_loss=0.03219, over 1971811.40 frames. ], batch size: 63, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:56:55,446 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=143428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:57:08,204 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.099e+02 2.101e+02 2.553e+02 3.059e+02 5.821e+02, threshold=5.106e+02, percent-clipped=2.0 2022-12-08 15:57:24,956 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143461.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:57:35,011 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143473.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 15:58:04,931 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0089, 2.0348, 2.1210, 2.0764, 2.0433, 1.6484, 1.3119, 1.8592], device='cuda:1'), covar=tensor([0.0751, 0.0578, 0.0450, 0.0471, 0.0477, 0.1572, 0.2302, 0.0519], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0175, 0.0147, 0.0149, 0.0209, 0.0143, 0.0157, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 15:58:13,338 INFO [train.py:873] (1/4) Epoch 19, batch 7400, loss[loss=0.08743, simple_loss=0.1324, pruned_loss=0.02122, over 14516.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1395, pruned_loss=0.0329, over 1899359.03 frames. ], batch size: 43, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:58:29,177 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143534.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:58:36,573 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 2.037e+02 2.578e+02 3.136e+02 8.501e+02, threshold=5.155e+02, percent-clipped=4.0 2022-12-08 15:58:45,916 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6467, 1.6135, 2.7938, 1.6576, 2.7983, 2.7597, 2.0026, 2.9312], device='cuda:1'), covar=tensor([0.0285, 0.2489, 0.0440, 0.1820, 0.0465, 0.0536, 0.1345, 0.0301], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0156, 0.0161, 0.0169, 0.0167, 0.0179, 0.0134, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 15:59:41,170 INFO [train.py:873] (1/4) Epoch 19, batch 7500, loss[loss=0.1027, simple_loss=0.1211, pruned_loss=0.04216, over 2645.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.1395, pruned_loss=0.03245, over 1954976.25 frames. ], batch size: 100, lr: 4.07e-03, grad_scale: 8.0 2022-12-08 15:59:46,307 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9099, 3.6886, 3.6566, 4.0102, 3.6986, 3.5361, 4.0634, 3.3857], device='cuda:1'), covar=tensor([0.0622, 0.1099, 0.0509, 0.0502, 0.0870, 0.1455, 0.0551, 0.0606], device='cuda:1'), in_proj_covar=tensor([0.0183, 0.0279, 0.0205, 0.0202, 0.0188, 0.0162, 0.0294, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 16:00:00,459 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 16:00:03,425 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.124e+02 2.640e+02 3.469e+02 7.311e+02, threshold=5.280e+02, percent-clipped=6.0 2022-12-08 16:00:04,582 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4380, 2.5897, 1.7292, 2.2331, 2.2784, 2.1266, 1.9653, 2.1665], device='cuda:1'), covar=tensor([0.0311, 0.0265, 0.0399, 0.0266, 0.0244, 0.0443, 0.0532, 0.0496], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0022, 0.0035, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 16:00:10,165 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9165, 2.7501, 2.7664, 2.9525, 2.8021, 2.8450, 2.9929, 2.4969], device='cuda:1'), covar=tensor([0.0721, 0.1185, 0.0641, 0.0611, 0.0875, 0.0560, 0.0718, 0.0702], device='cuda:1'), in_proj_covar=tensor([0.0182, 0.0278, 0.0205, 0.0202, 0.0187, 0.0162, 0.0294, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 16:01:11,360 INFO [train.py:873] (1/4) Epoch 20, batch 0, loss[loss=0.1035, simple_loss=0.1298, pruned_loss=0.03859, over 3908.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.1298, pruned_loss=0.03859, over 3908.00 frames. ], batch size: 100, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:01:11,360 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 16:01:17,456 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5961, 2.6847, 2.5045, 2.7139, 2.2289, 2.6723, 2.6805, 2.6130], device='cuda:1'), covar=tensor([0.0846, 0.1085, 0.0964, 0.1063, 0.1340, 0.0889, 0.0922, 0.1215], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0146, 0.0149, 0.0162, 0.0150, 0.0126, 0.0169, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:01:18,885 INFO [train.py:905] (1/4) Epoch 20, validation: loss=0.1452, simple_loss=0.1824, pruned_loss=0.05396, over 857387.00 frames. 2022-12-08 16:01:18,886 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 16:02:15,749 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.236e+01 1.946e+02 2.466e+02 3.510e+02 7.205e+02, threshold=4.933e+02, percent-clipped=3.0 2022-12-08 16:02:22,543 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1204, 3.1033, 2.9273, 3.2378, 2.8480, 2.9369, 3.2026, 3.1090], device='cuda:1'), covar=tensor([0.0680, 0.1051, 0.1033, 0.0631, 0.1262, 0.0788, 0.0701, 0.0756], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0163, 0.0150, 0.0126, 0.0169, 0.0150], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:02:27,765 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143756.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:02:40,088 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-08 16:02:47,466 INFO [train.py:873] (1/4) Epoch 20, batch 100, loss[loss=0.1131, simple_loss=0.1222, pruned_loss=0.052, over 1220.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.1401, pruned_loss=0.03286, over 836217.84 frames. ], batch size: 100, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:03:05,085 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 16:03:31,970 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143829.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:03:34,770 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7912, 2.3561, 3.6478, 2.7576, 3.6896, 3.5136, 3.3956, 3.0213], device='cuda:1'), covar=tensor([0.0914, 0.2998, 0.1018, 0.1819, 0.0799, 0.1119, 0.1263, 0.1628], device='cuda:1'), in_proj_covar=tensor([0.0351, 0.0310, 0.0389, 0.0301, 0.0367, 0.0324, 0.0361, 0.0298], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:03:43,924 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.111e+02 2.727e+02 3.354e+02 8.934e+02, threshold=5.454e+02, percent-clipped=6.0 2022-12-08 16:04:15,259 INFO [train.py:873] (1/4) Epoch 20, batch 200, loss[loss=0.1017, simple_loss=0.1212, pruned_loss=0.04108, over 2608.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1395, pruned_loss=0.03288, over 1277320.42 frames. ], batch size: 100, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:05:11,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.198e+02 1.866e+02 2.261e+02 3.101e+02 5.918e+02, threshold=4.522e+02, percent-clipped=1.0 2022-12-08 16:05:32,946 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5847, 3.9325, 3.7275, 3.6425, 2.7738, 3.8746, 3.7512, 2.2486], device='cuda:1'), covar=tensor([0.1133, 0.0863, 0.0937, 0.0777, 0.0885, 0.0444, 0.0613, 0.1533], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0093, 0.0072, 0.0078, 0.0101, 0.0092, 0.0103, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:05:43,233 INFO [train.py:873] (1/4) Epoch 20, batch 300, loss[loss=0.1105, simple_loss=0.1478, pruned_loss=0.03659, over 13522.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1398, pruned_loss=0.03286, over 1511718.46 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:05:56,780 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7848, 1.2906, 2.5062, 2.2033, 2.3494, 2.5623, 1.6436, 2.5050], device='cuda:1'), covar=tensor([0.1075, 0.1507, 0.0282, 0.0565, 0.0606, 0.0287, 0.0957, 0.0333], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0133, 0.0171, 0.0150, 0.0144, 0.0127, 0.0126], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 16:06:39,226 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.977e+01 2.046e+02 2.429e+02 3.312e+02 9.305e+02, threshold=4.857e+02, percent-clipped=8.0 2022-12-08 16:06:43,504 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8024, 3.9423, 4.1656, 3.6133, 3.9805, 4.0832, 1.5814, 3.8205], device='cuda:1'), covar=tensor([0.0390, 0.0371, 0.0338, 0.0545, 0.0319, 0.0322, 0.3183, 0.0274], device='cuda:1'), in_proj_covar=tensor([0.0176, 0.0176, 0.0148, 0.0150, 0.0210, 0.0143, 0.0158, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 16:06:50,873 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144056.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:06:55,925 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9991, 3.7885, 3.4889, 3.7154, 3.8878, 3.9319, 3.9899, 3.9763], device='cuda:1'), covar=tensor([0.0791, 0.0546, 0.2056, 0.2079, 0.0758, 0.0804, 0.0811, 0.0772], device='cuda:1'), in_proj_covar=tensor([0.0399, 0.0282, 0.0458, 0.0580, 0.0361, 0.0468, 0.0400, 0.0403], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:07:10,842 INFO [train.py:873] (1/4) Epoch 20, batch 400, loss[loss=0.07801, simple_loss=0.1206, pruned_loss=0.0177, over 13987.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1386, pruned_loss=0.03178, over 1689668.27 frames. ], batch size: 20, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:07:32,242 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=144104.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:07:33,147 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9365, 1.5274, 1.9602, 1.3270, 1.7281, 2.0234, 1.8115, 1.7686], device='cuda:1'), covar=tensor([0.0877, 0.0731, 0.0798, 0.1172, 0.1413, 0.0901, 0.0833, 0.1532], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0170, 0.0142, 0.0125, 0.0146, 0.0157, 0.0140, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:07:54,121 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144129.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:07:55,310 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2022-12-08 16:07:55,951 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9007, 1.9103, 2.0278, 1.9791, 1.9524, 1.8320, 1.8612, 1.3215], device='cuda:1'), covar=tensor([0.0190, 0.0356, 0.0235, 0.0242, 0.0177, 0.0265, 0.0258, 0.0385], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 16:08:07,211 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.224e+02 2.595e+02 3.172e+02 7.454e+02, threshold=5.190e+02, percent-clipped=6.0 2022-12-08 16:08:35,921 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=144177.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:08:37,519 INFO [train.py:873] (1/4) Epoch 20, batch 500, loss[loss=0.1116, simple_loss=0.1546, pruned_loss=0.0343, over 13810.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1395, pruned_loss=0.03282, over 1823080.74 frames. ], batch size: 23, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:09:34,775 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.083e+02 2.620e+02 3.196e+02 5.834e+02, threshold=5.241e+02, percent-clipped=3.0 2022-12-08 16:09:53,121 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9682, 2.5580, 3.8388, 2.9752, 3.8389, 3.7107, 3.6333, 3.2172], device='cuda:1'), covar=tensor([0.0821, 0.2754, 0.0925, 0.1548, 0.0826, 0.0970, 0.1261, 0.1794], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0306, 0.0387, 0.0298, 0.0365, 0.0322, 0.0360, 0.0297], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:10:02,291 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-08 16:10:05,212 INFO [train.py:873] (1/4) Epoch 20, batch 600, loss[loss=0.1342, simple_loss=0.1599, pruned_loss=0.05421, over 10345.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1397, pruned_loss=0.03261, over 1915334.95 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:10:08,662 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9170, 2.8409, 2.4561, 2.5643, 2.9272, 2.8840, 2.8912, 2.9196], device='cuda:1'), covar=tensor([0.1386, 0.0942, 0.3196, 0.3677, 0.1237, 0.1661, 0.1819, 0.1248], device='cuda:1'), in_proj_covar=tensor([0.0396, 0.0280, 0.0455, 0.0576, 0.0358, 0.0465, 0.0398, 0.0399], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:10:28,509 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 16:10:41,972 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2022-12-08 16:10:58,345 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4358, 1.3712, 2.7021, 1.5280, 2.5478, 2.6614, 1.9820, 2.6647], device='cuda:1'), covar=tensor([0.0507, 0.3305, 0.0596, 0.2385, 0.0667, 0.0680, 0.1302, 0.0595], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0156, 0.0162, 0.0169, 0.0166, 0.0180, 0.0133, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:11:02,325 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.036e+02 2.548e+02 2.967e+02 5.900e+02, threshold=5.097e+02, percent-clipped=2.0 2022-12-08 16:11:33,547 INFO [train.py:873] (1/4) Epoch 20, batch 700, loss[loss=0.1054, simple_loss=0.1449, pruned_loss=0.03294, over 14299.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1392, pruned_loss=0.03217, over 1912853.31 frames. ], batch size: 37, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:11:33,633 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4788, 5.4093, 5.0478, 5.5062, 5.1226, 5.1281, 5.5788, 5.1932], device='cuda:1'), covar=tensor([0.0550, 0.0678, 0.0745, 0.0492, 0.0662, 0.0404, 0.0451, 0.0684], device='cuda:1'), in_proj_covar=tensor([0.0150, 0.0149, 0.0151, 0.0166, 0.0153, 0.0129, 0.0172, 0.0152], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:11:49,093 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5603, 2.7099, 2.7890, 2.7364, 2.7442, 2.5025, 1.6189, 2.5350], device='cuda:1'), covar=tensor([0.0589, 0.0475, 0.0452, 0.0442, 0.0385, 0.0959, 0.2649, 0.0411], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0174, 0.0146, 0.0149, 0.0208, 0.0142, 0.0157, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 16:12:02,241 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144411.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:12:31,474 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.983e+02 2.498e+02 3.051e+02 6.384e+02, threshold=4.995e+02, percent-clipped=1.0 2022-12-08 16:12:46,071 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.5027, 1.6823, 1.9346, 1.8889, 1.7995, 1.8398, 1.5973, 1.4119], device='cuda:1'), covar=tensor([0.1038, 0.1605, 0.0491, 0.0646, 0.1172, 0.0942, 0.1468, 0.1779], device='cuda:1'), in_proj_covar=tensor([0.0137, 0.0094, 0.0073, 0.0079, 0.0103, 0.0094, 0.0104, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 16:12:47,090 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6080, 2.0087, 3.6818, 2.6114, 3.5912, 1.9454, 2.7761, 3.6201], device='cuda:1'), covar=tensor([0.0697, 0.3653, 0.0568, 0.4923, 0.0654, 0.3024, 0.1358, 0.0631], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0198, 0.0221, 0.0266, 0.0239, 0.0201, 0.0202, 0.0219], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 16:12:56,230 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:13:02,249 INFO [train.py:873] (1/4) Epoch 20, batch 800, loss[loss=0.1364, simple_loss=0.1315, pruned_loss=0.07066, over 1279.00 frames. ], tot_loss[loss=0.1019, simple_loss=0.1394, pruned_loss=0.0322, over 1936179.31 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:13:58,405 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.092e+02 2.630e+02 3.168e+02 5.138e+02, threshold=5.259e+02, percent-clipped=1.0 2022-12-08 16:14:05,703 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2022-12-08 16:14:25,407 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.3665, 0.9282, 1.2034, 0.8004, 1.1251, 1.3549, 1.0755, 1.0824], device='cuda:1'), covar=tensor([0.0468, 0.1097, 0.0851, 0.0600, 0.1307, 0.0968, 0.0672, 0.1376], device='cuda:1'), in_proj_covar=tensor([0.0158, 0.0172, 0.0143, 0.0127, 0.0148, 0.0159, 0.0142, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0007, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:14:29,535 INFO [train.py:873] (1/4) Epoch 20, batch 900, loss[loss=0.1256, simple_loss=0.1566, pruned_loss=0.04726, over 14182.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.1394, pruned_loss=0.03257, over 1935270.64 frames. ], batch size: 89, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:14:43,280 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7886, 3.6099, 3.3454, 3.4833, 3.6992, 3.7098, 3.7332, 3.7609], device='cuda:1'), covar=tensor([0.0693, 0.0580, 0.1959, 0.2294, 0.0677, 0.0819, 0.0948, 0.0729], device='cuda:1'), in_proj_covar=tensor([0.0394, 0.0280, 0.0453, 0.0573, 0.0356, 0.0461, 0.0395, 0.0397], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:15:27,786 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 2.044e+02 2.378e+02 2.904e+02 6.661e+02, threshold=4.757e+02, percent-clipped=2.0 2022-12-08 16:15:51,063 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9027, 1.8184, 3.0948, 2.1823, 2.9848, 1.8739, 2.3499, 2.9679], device='cuda:1'), covar=tensor([0.1153, 0.3683, 0.0703, 0.4498, 0.1032, 0.2926, 0.1411, 0.0813], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0198, 0.0221, 0.0266, 0.0239, 0.0201, 0.0202, 0.0219], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 16:15:58,943 INFO [train.py:873] (1/4) Epoch 20, batch 1000, loss[loss=0.1131, simple_loss=0.1478, pruned_loss=0.03918, over 14264.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1395, pruned_loss=0.03271, over 1936686.93 frames. ], batch size: 44, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:16:14,205 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7043, 3.4809, 3.1982, 2.3100, 3.1390, 3.4220, 3.6956, 3.0347], device='cuda:1'), covar=tensor([0.0596, 0.0860, 0.0937, 0.1259, 0.0938, 0.0674, 0.0812, 0.1054], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0171, 0.0143, 0.0127, 0.0147, 0.0158, 0.0142, 0.0146], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:16:56,767 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 1.902e+02 2.363e+02 2.821e+02 5.781e+02, threshold=4.726e+02, percent-clipped=3.0 2022-12-08 16:17:01,905 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2022-12-08 16:17:17,191 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=144767.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:17:27,640 INFO [train.py:873] (1/4) Epoch 20, batch 1100, loss[loss=0.1179, simple_loss=0.1282, pruned_loss=0.05378, over 2635.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1385, pruned_loss=0.03226, over 1985816.21 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:17:32,116 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3147, 1.9168, 2.2798, 1.6103, 1.9818, 2.3035, 2.2224, 2.0629], device='cuda:1'), covar=tensor([0.0873, 0.0551, 0.0952, 0.1316, 0.1309, 0.0841, 0.0796, 0.1344], device='cuda:1'), in_proj_covar=tensor([0.0157, 0.0171, 0.0143, 0.0126, 0.0147, 0.0158, 0.0141, 0.0145], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:18:25,976 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.171e+02 2.584e+02 3.315e+02 6.802e+02, threshold=5.169e+02, percent-clipped=3.0 2022-12-08 16:18:43,728 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.0685, 2.4020, 5.0835, 3.3920, 4.7419, 2.2562, 3.6019, 4.8388], device='cuda:1'), covar=tensor([0.0364, 0.3660, 0.0255, 0.5542, 0.0492, 0.3060, 0.1375, 0.0328], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0197, 0.0221, 0.0265, 0.0239, 0.0200, 0.0202, 0.0219], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 16:18:56,028 INFO [train.py:873] (1/4) Epoch 20, batch 1200, loss[loss=0.1565, simple_loss=0.1554, pruned_loss=0.07879, over 1279.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.1392, pruned_loss=0.03258, over 1979697.90 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:19:15,248 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.9381, 3.3796, 2.6436, 4.2277, 3.9798, 4.0557, 3.5412, 2.7056], device='cuda:1'), covar=tensor([0.0782, 0.1362, 0.3318, 0.0577, 0.0908, 0.1038, 0.1193, 0.3336], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0289, 0.0260, 0.0293, 0.0324, 0.0304, 0.0258, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:19:54,168 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 1.946e+02 2.467e+02 2.965e+02 6.907e+02, threshold=4.934e+02, percent-clipped=1.0 2022-12-08 16:19:59,984 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7429, 3.3551, 2.6319, 3.8399, 3.7229, 3.7343, 3.2867, 2.6528], device='cuda:1'), covar=tensor([0.0813, 0.1268, 0.3077, 0.0707, 0.0891, 0.0995, 0.1240, 0.2784], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0287, 0.0258, 0.0291, 0.0322, 0.0303, 0.0257, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:20:23,993 INFO [train.py:873] (1/4) Epoch 20, batch 1300, loss[loss=0.08814, simple_loss=0.1384, pruned_loss=0.01895, over 14543.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1392, pruned_loss=0.03284, over 1966481.47 frames. ], batch size: 43, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:20:39,325 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144996.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:20:42,243 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2022-12-08 16:21:10,368 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:26,336 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.036e+02 2.136e+02 2.595e+02 3.096e+02 6.124e+02, threshold=5.190e+02, percent-clipped=3.0 2022-12-08 16:21:36,818 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145057.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:21:38,378 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0789, 2.0078, 1.6590, 1.7161, 2.0294, 2.0476, 2.0174, 1.9947], device='cuda:1'), covar=tensor([0.1465, 0.1060, 0.3208, 0.3339, 0.1572, 0.1652, 0.1946, 0.1604], device='cuda:1'), in_proj_covar=tensor([0.0396, 0.0280, 0.0455, 0.0574, 0.0356, 0.0463, 0.0395, 0.0398], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:21:44,327 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:45,146 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145067.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:55,279 INFO [train.py:873] (1/4) Epoch 20, batch 1400, loss[loss=0.08966, simple_loss=0.1339, pruned_loss=0.02271, over 14260.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1395, pruned_loss=0.03273, over 1916544.98 frames. ], batch size: 28, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:22:03,055 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:26,071 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:36,148 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:52,062 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.811e+01 2.056e+02 2.726e+02 3.195e+02 8.712e+02, threshold=5.453e+02, percent-clipped=3.0 2022-12-08 16:22:54,940 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.6468, 1.6594, 1.7651, 1.8467, 1.7828, 1.2133, 1.5261, 1.5929], device='cuda:1'), covar=tensor([0.0827, 0.0867, 0.0597, 0.0723, 0.0574, 0.0881, 0.0679, 0.0584], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 16:23:20,738 INFO [train.py:873] (1/4) Epoch 20, batch 1500, loss[loss=0.1498, simple_loss=0.1396, pruned_loss=0.08005, over 1208.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.139, pruned_loss=0.03228, over 1959923.80 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 4.0 2022-12-08 16:23:23,485 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2022-12-08 16:23:29,260 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.1719, 5.0346, 4.5442, 4.8506, 4.8501, 5.0941, 5.2393, 5.1846], device='cuda:1'), covar=tensor([0.0791, 0.0369, 0.2222, 0.2217, 0.0797, 0.0763, 0.0662, 0.0692], device='cuda:1'), in_proj_covar=tensor([0.0394, 0.0280, 0.0455, 0.0575, 0.0357, 0.0462, 0.0396, 0.0398], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:23:35,373 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9186, 1.7794, 4.2003, 3.9325, 3.8956, 4.3369, 3.6386, 4.2605], device='cuda:1'), covar=tensor([0.1719, 0.1704, 0.0147, 0.0273, 0.0285, 0.0165, 0.0287, 0.0160], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0156, 0.0131, 0.0169, 0.0149, 0.0143, 0.0126, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 16:23:43,592 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4117, 2.0606, 2.4591, 2.5637, 2.3725, 2.0576, 2.5043, 2.1812], device='cuda:1'), covar=tensor([0.0577, 0.1308, 0.0706, 0.0537, 0.0798, 0.1466, 0.0619, 0.0859], device='cuda:1'), in_proj_covar=tensor([0.0294, 0.0261, 0.0375, 0.0330, 0.0271, 0.0308, 0.0314, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:23:45,487 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.72 vs. limit=5.0 2022-12-08 16:24:05,581 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2022-12-08 16:24:19,275 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.332e+01 2.024e+02 2.567e+02 3.376e+02 1.316e+03, threshold=5.133e+02, percent-clipped=5.0 2022-12-08 16:24:48,111 INFO [train.py:873] (1/4) Epoch 20, batch 1600, loss[loss=0.08424, simple_loss=0.1248, pruned_loss=0.02186, over 14244.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1388, pruned_loss=0.03215, over 2000464.53 frames. ], batch size: 25, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:25:05,345 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-08 16:25:09,216 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2216, 1.5808, 2.4044, 2.0224, 2.2065, 1.6493, 2.0340, 2.2440], device='cuda:1'), covar=tensor([0.2585, 0.4097, 0.0847, 0.2151, 0.1629, 0.2452, 0.1112, 0.1320], device='cuda:1'), in_proj_covar=tensor([0.0256, 0.0200, 0.0223, 0.0267, 0.0242, 0.0204, 0.0204, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 16:25:14,421 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7123, 2.4342, 3.7126, 3.8429, 3.6231, 2.2318, 3.7310, 2.7419], device='cuda:1'), covar=tensor([0.0574, 0.1343, 0.0918, 0.0616, 0.0719, 0.2298, 0.0650, 0.1313], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0261, 0.0375, 0.0331, 0.0271, 0.0308, 0.0315, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:25:46,272 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 1.921e+02 2.439e+02 2.878e+02 7.278e+02, threshold=4.878e+02, percent-clipped=1.0 2022-12-08 16:25:51,940 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145352.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 16:25:53,013 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5077, 2.7202, 2.8606, 2.5847, 2.5208, 2.1575, 2.2715, 2.2173], device='cuda:1'), covar=tensor([0.0328, 0.0249, 0.0188, 0.0265, 0.0236, 0.0416, 0.0466, 0.0651], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0023, 0.0023, 0.0035, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 16:26:08,354 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:26:08,640 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 16:26:12,485 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.4153, 5.1533, 4.6789, 5.0203, 4.9422, 5.2797, 5.3430, 5.3421], device='cuda:1'), covar=tensor([0.0606, 0.0334, 0.1920, 0.2057, 0.0727, 0.0714, 0.0725, 0.0652], device='cuda:1'), in_proj_covar=tensor([0.0397, 0.0281, 0.0457, 0.0577, 0.0359, 0.0466, 0.0400, 0.0400], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:26:15,270 INFO [train.py:873] (1/4) Epoch 20, batch 1700, loss[loss=0.1436, simple_loss=0.1388, pruned_loss=0.07419, over 1198.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1385, pruned_loss=0.03183, over 1986595.99 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:26:18,712 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:26:52,680 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145422.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:27:01,930 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:27:02,772 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8981, 1.6088, 3.5973, 3.3918, 3.4799, 3.7332, 2.8661, 3.6457], device='cuda:1'), covar=tensor([0.1965, 0.1966, 0.0233, 0.0340, 0.0364, 0.0216, 0.0380, 0.0214], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0158, 0.0132, 0.0171, 0.0151, 0.0144, 0.0127, 0.0127], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 16:27:13,650 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 16:27:13,922 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.159e+01 2.091e+02 2.554e+02 3.165e+02 6.798e+02, threshold=5.109e+02, percent-clipped=6.0 2022-12-08 16:27:42,938 INFO [train.py:873] (1/4) Epoch 20, batch 1800, loss[loss=0.1054, simple_loss=0.1434, pruned_loss=0.03369, over 14380.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1392, pruned_loss=0.03235, over 1994851.64 frames. ], batch size: 73, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:28:41,420 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.270e+02 2.808e+02 3.331e+02 1.195e+03, threshold=5.616e+02, percent-clipped=5.0 2022-12-08 16:29:10,284 INFO [train.py:873] (1/4) Epoch 20, batch 1900, loss[loss=0.0952, simple_loss=0.1364, pruned_loss=0.027, over 14366.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1393, pruned_loss=0.03278, over 1855085.05 frames. ], batch size: 73, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:29:23,567 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:29:40,074 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3122, 3.0807, 3.0573, 3.2810, 3.1031, 3.2365, 3.3493, 2.7831], device='cuda:1'), covar=tensor([0.0543, 0.1021, 0.0574, 0.0523, 0.0794, 0.0447, 0.0581, 0.0632], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0272, 0.0200, 0.0200, 0.0184, 0.0160, 0.0288, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 16:29:53,277 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4148, 1.4383, 3.4008, 1.5599, 3.2426, 3.5268, 2.4702, 3.7216], device='cuda:1'), covar=tensor([0.0265, 0.3271, 0.0472, 0.2553, 0.0916, 0.0401, 0.0929, 0.0214], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0157, 0.0163, 0.0170, 0.0168, 0.0182, 0.0134, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:30:07,091 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:30:09,685 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.010e+02 2.453e+02 2.850e+02 4.730e+02, threshold=4.905e+02, percent-clipped=0.0 2022-12-08 16:30:15,033 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145652.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 16:30:17,606 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145655.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:30:38,452 INFO [train.py:873] (1/4) Epoch 20, batch 2000, loss[loss=0.08837, simple_loss=0.1355, pruned_loss=0.02064, over 14162.00 frames. ], tot_loss[loss=0.1019, simple_loss=0.1393, pruned_loss=0.03221, over 1906198.78 frames. ], batch size: 35, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:30:42,045 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145683.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:30:47,597 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8875, 2.5108, 3.7486, 2.7976, 3.7766, 3.6582, 3.5141, 3.1090], device='cuda:1'), covar=tensor([0.0900, 0.2910, 0.1166, 0.1814, 0.0917, 0.1007, 0.1384, 0.1667], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0307, 0.0388, 0.0298, 0.0364, 0.0319, 0.0358, 0.0295], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:30:56,733 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145700.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:00,309 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:16,150 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145722.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:20,721 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145727.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:23,995 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:37,254 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.155e+02 1.878e+02 2.490e+02 3.077e+02 5.719e+02, threshold=4.981e+02, percent-clipped=2.0 2022-12-08 16:31:58,235 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:32:05,944 INFO [train.py:873] (1/4) Epoch 20, batch 2100, loss[loss=0.1321, simple_loss=0.1531, pruned_loss=0.05552, over 7764.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1386, pruned_loss=0.03182, over 1889107.91 frames. ], batch size: 100, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:32:43,321 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145821.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:32:52,965 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2311, 4.1215, 3.6562, 3.8202, 4.1081, 4.1693, 4.3092, 4.2293], device='cuda:1'), covar=tensor([0.1167, 0.0628, 0.2526, 0.3116, 0.1044, 0.1273, 0.1096, 0.1143], device='cuda:1'), in_proj_covar=tensor([0.0397, 0.0281, 0.0455, 0.0579, 0.0359, 0.0467, 0.0400, 0.0399], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:33:04,938 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.012e+02 2.500e+02 3.223e+02 4.888e+02, threshold=5.000e+02, percent-clipped=0.0 2022-12-08 16:33:33,924 INFO [train.py:873] (1/4) Epoch 20, batch 2200, loss[loss=0.1003, simple_loss=0.1406, pruned_loss=0.03006, over 14136.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1391, pruned_loss=0.03261, over 1884444.09 frames. ], batch size: 29, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:33:36,662 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145882.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:34:08,960 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6148, 1.4769, 2.7867, 1.5536, 2.8035, 2.7779, 2.1196, 2.9367], device='cuda:1'), covar=tensor([0.0288, 0.2891, 0.0463, 0.2042, 0.0461, 0.0553, 0.1066, 0.0283], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0157, 0.0163, 0.0170, 0.0168, 0.0182, 0.0134, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:34:32,821 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.107e+02 2.702e+02 3.321e+02 7.589e+02, threshold=5.404e+02, percent-clipped=7.0 2022-12-08 16:34:36,629 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145950.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:34:59,898 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3672, 3.8056, 3.5588, 3.5731, 2.7501, 3.7173, 3.5293, 2.0269], device='cuda:1'), covar=tensor([0.1137, 0.0575, 0.0955, 0.0708, 0.0816, 0.0405, 0.0831, 0.1620], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 16:35:01,455 INFO [train.py:873] (1/4) Epoch 20, batch 2300, loss[loss=0.1474, simple_loss=0.1535, pruned_loss=0.07063, over 1313.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1381, pruned_loss=0.03254, over 1830473.89 frames. ], batch size: 100, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:35:19,647 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145999.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:35:44,240 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:36:00,973 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.169e+02 2.442e+02 2.958e+02 6.970e+02, threshold=4.884e+02, percent-clipped=2.0 2022-12-08 16:36:26,854 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:36:29,278 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 16:36:30,340 INFO [train.py:873] (1/4) Epoch 20, batch 2400, loss[loss=0.1436, simple_loss=0.1393, pruned_loss=0.07395, over 1291.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.139, pruned_loss=0.03278, over 1934225.77 frames. ], batch size: 100, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:36:32,634 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 16:36:53,902 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 16:37:27,524 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9987, 1.8808, 2.1411, 1.9741, 1.9783, 1.7411, 1.5600, 1.3459], device='cuda:1'), covar=tensor([0.0187, 0.0377, 0.0265, 0.0351, 0.0242, 0.0325, 0.0274, 0.0442], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 16:37:29,040 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 1.997e+02 2.339e+02 2.913e+02 8.016e+02, threshold=4.679e+02, percent-clipped=1.0 2022-12-08 16:37:30,614 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2022-12-08 16:37:31,922 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146149.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:37:47,407 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 16:37:56,527 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146177.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:37:58,096 INFO [train.py:873] (1/4) Epoch 20, batch 2500, loss[loss=0.1476, simple_loss=0.1401, pruned_loss=0.07752, over 1304.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.1383, pruned_loss=0.03212, over 1926157.43 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:38:26,578 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146210.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:38:58,168 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.933e+01 2.087e+02 2.444e+02 3.038e+02 5.934e+02, threshold=4.888e+02, percent-clipped=3.0 2022-12-08 16:39:01,845 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146250.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:39:15,773 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.4443, 2.0922, 2.1428, 2.4652, 2.3459, 2.0085, 1.7293, 1.8371], device='cuda:1'), covar=tensor([0.0197, 0.0394, 0.0449, 0.0408, 0.0212, 0.0510, 0.0487, 0.0529], device='cuda:1'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 16:39:27,231 INFO [train.py:873] (1/4) Epoch 20, batch 2600, loss[loss=0.08016, simple_loss=0.1222, pruned_loss=0.01906, over 13954.00 frames. ], tot_loss[loss=0.1007, simple_loss=0.1379, pruned_loss=0.0318, over 1895075.49 frames. ], batch size: 19, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:39:43,722 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146298.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:39:44,677 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:40:25,673 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 8.546e+01 2.007e+02 2.593e+02 2.996e+02 8.778e+02, threshold=5.186e+02, percent-clipped=5.0 2022-12-08 16:40:26,674 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:40:54,845 INFO [train.py:873] (1/4) Epoch 20, batch 2700, loss[loss=0.145, simple_loss=0.1436, pruned_loss=0.07314, over 1267.00 frames. ], tot_loss[loss=0.1017, simple_loss=0.1389, pruned_loss=0.03224, over 1938848.87 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:41:53,288 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.024e+02 2.634e+02 3.181e+02 1.190e+03, threshold=5.268e+02, percent-clipped=3.0 2022-12-08 16:42:20,415 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146477.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:42:21,925 INFO [train.py:873] (1/4) Epoch 20, batch 2800, loss[loss=0.1045, simple_loss=0.1441, pruned_loss=0.03243, over 14241.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.1387, pruned_loss=0.03198, over 1933177.80 frames. ], batch size: 69, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:42:39,714 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3979, 1.9754, 2.1606, 2.3327, 2.2074, 1.3910, 2.1355, 2.3588], device='cuda:1'), covar=tensor([0.0866, 0.0987, 0.1049, 0.1088, 0.0780, 0.0914, 0.0757, 0.0684], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0038, 0.0042, 0.0035, 0.0037, 0.0052, 0.0040, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 16:42:44,614 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:43:02,235 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146525.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:43:20,289 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 1.986e+02 2.533e+02 3.212e+02 8.104e+02, threshold=5.065e+02, percent-clipped=2.0 2022-12-08 16:43:48,732 INFO [train.py:873] (1/4) Epoch 20, batch 2900, loss[loss=0.09218, simple_loss=0.1367, pruned_loss=0.02382, over 14021.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.1389, pruned_loss=0.03186, over 1967200.56 frames. ], batch size: 26, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:43:53,352 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8959, 1.4918, 3.0153, 2.6867, 2.8505, 3.0186, 2.2215, 2.9924], device='cuda:1'), covar=tensor([0.1258, 0.1574, 0.0194, 0.0431, 0.0420, 0.0221, 0.0574, 0.0209], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0131, 0.0170, 0.0149, 0.0143, 0.0125, 0.0125], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 16:43:59,383 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:44:10,361 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3945, 3.1080, 3.9328, 2.8891, 2.4457, 3.0425, 1.8972, 3.4319], device='cuda:1'), covar=tensor([0.0757, 0.0929, 0.0399, 0.1864, 0.1731, 0.1066, 0.2892, 0.0849], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0103, 0.0097, 0.0102, 0.0115, 0.0093, 0.0117, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 16:44:30,216 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:44:31,021 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8336, 0.8241, 0.7563, 0.7815, 0.8297, 0.4812, 0.8056, 0.8426], device='cuda:1'), covar=tensor([0.0431, 0.0551, 0.0804, 0.0565, 0.0365, 0.0407, 0.1173, 0.0950], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 16:44:47,462 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.123e+02 2.667e+02 3.296e+02 6.992e+02, threshold=5.334e+02, percent-clipped=2.0 2022-12-08 16:44:52,704 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146652.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:45:02,080 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2022-12-08 16:45:16,678 INFO [train.py:873] (1/4) Epoch 20, batch 3000, loss[loss=0.09901, simple_loss=0.1417, pruned_loss=0.02814, over 14269.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.1386, pruned_loss=0.03199, over 1876938.66 frames. ], batch size: 60, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:45:16,678 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 16:45:25,136 INFO [train.py:905] (1/4) Epoch 20, validation: loss=0.1444, simple_loss=0.1794, pruned_loss=0.05469, over 857387.00 frames. 2022-12-08 16:45:25,136 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 16:45:32,057 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146687.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:45:39,062 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7782, 1.7382, 1.7963, 1.8994, 1.8658, 1.2278, 1.5087, 1.7265], device='cuda:1'), covar=tensor([0.0587, 0.0662, 0.0590, 0.0501, 0.0449, 0.0898, 0.0893, 0.0633], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0038, 0.0042, 0.0034, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 16:46:23,235 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.030e+02 2.519e+02 3.160e+02 5.604e+02, threshold=5.039e+02, percent-clipped=1.0 2022-12-08 16:46:33,137 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-08 16:46:34,896 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 16:46:52,204 INFO [train.py:873] (1/4) Epoch 20, batch 3100, loss[loss=0.0912, simple_loss=0.1368, pruned_loss=0.02282, over 14537.00 frames. ], tot_loss[loss=0.1009, simple_loss=0.1385, pruned_loss=0.03163, over 1927691.23 frames. ], batch size: 49, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:47:14,827 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146805.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:47:19,124 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6470, 1.4775, 2.7800, 1.5617, 2.7753, 2.7657, 2.1050, 2.9140], device='cuda:1'), covar=tensor([0.0286, 0.2751, 0.0428, 0.1922, 0.0420, 0.0529, 0.1090, 0.0288], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0157, 0.0163, 0.0170, 0.0168, 0.0181, 0.0134, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:47:48,895 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146844.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:47:50,640 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.017e+02 2.599e+02 3.064e+02 9.305e+02, threshold=5.198e+02, percent-clipped=5.0 2022-12-08 16:47:56,844 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146853.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:48:19,589 INFO [train.py:873] (1/4) Epoch 20, batch 3200, loss[loss=0.09511, simple_loss=0.1405, pruned_loss=0.02488, over 14215.00 frames. ], tot_loss[loss=0.1003, simple_loss=0.1384, pruned_loss=0.03116, over 1980562.03 frames. ], batch size: 35, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:48:43,347 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146905.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:49:18,825 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.042e+02 2.442e+02 3.021e+02 7.374e+02, threshold=4.884e+02, percent-clipped=4.0 2022-12-08 16:49:19,700 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146947.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:49:47,630 INFO [train.py:873] (1/4) Epoch 20, batch 3300, loss[loss=0.1018, simple_loss=0.1439, pruned_loss=0.02983, over 14173.00 frames. ], tot_loss[loss=0.1006, simple_loss=0.1385, pruned_loss=0.03137, over 1989903.58 frames. ], batch size: 89, lr: 3.92e-03, grad_scale: 8.0 2022-12-08 16:49:50,260 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146982.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:50:47,402 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.122e+02 2.526e+02 3.271e+02 7.839e+02, threshold=5.052e+02, percent-clipped=6.0 2022-12-08 16:50:50,927 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9447, 1.8269, 1.6991, 1.9374, 1.7705, 1.9238, 1.7631, 1.6057], device='cuda:1'), covar=tensor([0.1302, 0.1235, 0.2242, 0.1092, 0.1584, 0.0949, 0.2051, 0.1247], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0285, 0.0258, 0.0293, 0.0323, 0.0304, 0.0256, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:51:14,963 INFO [train.py:873] (1/4) Epoch 20, batch 3400, loss[loss=0.1219, simple_loss=0.1608, pruned_loss=0.04144, over 14247.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1393, pruned_loss=0.03186, over 1974668.32 frames. ], batch size: 80, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:52:16,822 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 1.939e+02 2.278e+02 2.850e+02 6.486e+02, threshold=4.556e+02, percent-clipped=1.0 2022-12-08 16:52:25,753 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4152, 4.2695, 4.0161, 4.4854, 4.0223, 3.7265, 4.4732, 4.2186], device='cuda:1'), covar=tensor([0.0555, 0.0821, 0.0842, 0.0510, 0.0771, 0.0697, 0.0520, 0.0747], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0166, 0.0152, 0.0128, 0.0173, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 16:52:45,345 INFO [train.py:873] (1/4) Epoch 20, batch 3500, loss[loss=0.09143, simple_loss=0.1351, pruned_loss=0.02386, over 14221.00 frames. ], tot_loss[loss=0.09986, simple_loss=0.1381, pruned_loss=0.03082, over 2033085.79 frames. ], batch size: 60, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:52:53,189 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1469, 3.9031, 3.8290, 4.2043, 3.9250, 3.7361, 4.2409, 3.5609], device='cuda:1'), covar=tensor([0.0479, 0.0881, 0.0482, 0.0443, 0.0777, 0.1226, 0.0488, 0.0556], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0277, 0.0205, 0.0202, 0.0188, 0.0162, 0.0291, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 16:53:03,636 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147200.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:53:39,359 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147241.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:53:44,666 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.013e+02 2.574e+02 3.119e+02 7.865e+02, threshold=5.147e+02, percent-clipped=11.0 2022-12-08 16:53:44,872 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:12,621 INFO [train.py:873] (1/4) Epoch 20, batch 3600, loss[loss=0.09891, simple_loss=0.1384, pruned_loss=0.02968, over 14266.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1385, pruned_loss=0.03151, over 2010514.03 frames. ], batch size: 25, lr: 3.92e-03, grad_scale: 8.0 2022-12-08 16:54:13,666 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2709, 3.0028, 2.3962, 3.3742, 3.1741, 3.2342, 2.8751, 2.4623], device='cuda:1'), covar=tensor([0.0793, 0.1361, 0.2779, 0.0709, 0.0976, 0.0972, 0.1275, 0.2620], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0284, 0.0256, 0.0291, 0.0321, 0.0304, 0.0254, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:54:15,316 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147282.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:26,929 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147295.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:34,001 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147302.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:40,961 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147310.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:58,649 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147330.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:55:14,167 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.052e+02 2.112e+02 2.582e+02 3.106e+02 5.993e+02, threshold=5.164e+02, percent-clipped=4.0 2022-12-08 16:55:34,729 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:55:41,608 INFO [train.py:873] (1/4) Epoch 20, batch 3700, loss[loss=0.08763, simple_loss=0.1329, pruned_loss=0.02115, over 14085.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1392, pruned_loss=0.03254, over 1972868.41 frames. ], batch size: 26, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:55:58,585 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9979, 1.5994, 1.9937, 1.3824, 1.7877, 2.1187, 1.9232, 1.8315], device='cuda:1'), covar=tensor([0.1011, 0.0724, 0.0926, 0.1588, 0.1586, 0.1061, 0.0986, 0.1603], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0168, 0.0141, 0.0125, 0.0146, 0.0156, 0.0140, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:56:00,757 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5047, 2.3151, 2.2081, 2.4231, 2.5262, 1.5808, 2.5086, 2.3559], device='cuda:1'), covar=tensor([0.1078, 0.0650, 0.1021, 0.1408, 0.0772, 0.0795, 0.0909, 0.1159], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0034, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 16:56:06,066 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:56:07,000 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3828, 2.9696, 2.9879, 2.1067, 2.7904, 3.0335, 3.2750, 2.7481], device='cuda:1'), covar=tensor([0.0656, 0.1054, 0.0876, 0.1341, 0.0984, 0.0719, 0.0811, 0.1123], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0168, 0.0141, 0.0126, 0.0146, 0.0156, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:56:28,185 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-08 16:56:29,297 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147434.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:56:30,435 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 16:56:33,717 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8264, 3.2011, 2.5618, 3.9462, 3.7830, 3.8273, 3.2734, 2.6592], device='cuda:1'), covar=tensor([0.0794, 0.1465, 0.3203, 0.0622, 0.0945, 0.1070, 0.1278, 0.3014], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0287, 0.0259, 0.0293, 0.0325, 0.0307, 0.0257, 0.0243], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:56:41,246 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.266e+02 2.765e+02 3.309e+02 4.888e+02, threshold=5.530e+02, percent-clipped=0.0 2022-12-08 16:56:53,499 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1325, 4.0385, 3.6940, 2.9932, 3.5439, 3.7655, 4.0934, 3.4835], device='cuda:1'), covar=tensor([0.0420, 0.0719, 0.0632, 0.0987, 0.0714, 0.0589, 0.0653, 0.0796], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0168, 0.0141, 0.0126, 0.0146, 0.0157, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:56:58,744 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:57:07,329 INFO [train.py:873] (1/4) Epoch 20, batch 3800, loss[loss=0.09004, simple_loss=0.1382, pruned_loss=0.02095, over 14566.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1392, pruned_loss=0.03254, over 1928713.90 frames. ], batch size: 34, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:57:19,823 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4511, 3.1055, 3.8819, 2.7421, 2.5147, 3.2192, 1.8079, 3.3958], device='cuda:1'), covar=tensor([0.0691, 0.1151, 0.0615, 0.1881, 0.1792, 0.1176, 0.3026, 0.1070], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0103, 0.0096, 0.0101, 0.0115, 0.0093, 0.0116, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 16:57:21,642 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:57:25,959 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147500.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:57:40,167 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2133, 3.7021, 2.8523, 4.5362, 4.2055, 4.3231, 3.8506, 2.9094], device='cuda:1'), covar=tensor([0.0729, 0.1101, 0.3048, 0.0421, 0.0712, 0.0995, 0.0959, 0.2946], device='cuda:1'), in_proj_covar=tensor([0.0281, 0.0284, 0.0256, 0.0290, 0.0321, 0.0303, 0.0254, 0.0240], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 16:58:07,480 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 9.824e+01 2.099e+02 2.660e+02 3.307e+02 5.236e+02, threshold=5.319e+02, percent-clipped=0.0 2022-12-08 16:58:07,569 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:58:34,639 INFO [train.py:873] (1/4) Epoch 20, batch 3900, loss[loss=0.1067, simple_loss=0.1407, pruned_loss=0.03636, over 11146.00 frames. ], tot_loss[loss=0.101, simple_loss=0.138, pruned_loss=0.03193, over 1966688.42 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:58:50,307 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147597.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:59:35,033 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 1.905e+02 2.298e+02 3.086e+02 1.551e+03, threshold=4.597e+02, percent-clipped=3.0 2022-12-08 16:59:42,110 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5799, 2.3112, 2.5028, 1.6543, 2.0881, 2.4657, 2.4935, 2.1966], device='cuda:1'), covar=tensor([0.0847, 0.0660, 0.0797, 0.1257, 0.1427, 0.0798, 0.0834, 0.1329], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0168, 0.0141, 0.0126, 0.0146, 0.0157, 0.0140, 0.0144], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 16:59:44,857 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:59:51,004 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147666.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:00:01,897 INFO [train.py:873] (1/4) Epoch 20, batch 4000, loss[loss=0.1362, simple_loss=0.1436, pruned_loss=0.06442, over 3884.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1383, pruned_loss=0.03191, over 1949229.40 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:00:35,008 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1237, 2.1394, 2.2417, 1.9024, 1.7755, 1.9198, 1.9940, 2.0880], device='cuda:1'), covar=tensor([0.0522, 0.0686, 0.0267, 0.0475, 0.0481, 0.0541, 0.0561, 0.0534], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 17:00:37,462 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:00:49,359 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.4846, 1.1766, 1.9716, 1.7665, 1.7732, 2.0285, 1.4022, 2.0513], device='cuda:1'), covar=tensor([0.1060, 0.1622, 0.0394, 0.0699, 0.0930, 0.0454, 0.1009, 0.0421], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0171, 0.0148, 0.0142, 0.0125, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 17:01:01,455 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.183e+02 2.642e+02 3.333e+02 1.015e+03, threshold=5.284e+02, percent-clipped=7.0 2022-12-08 17:01:05,662 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-08 17:01:14,028 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:01:27,735 INFO [train.py:873] (1/4) Epoch 20, batch 4100, loss[loss=0.1225, simple_loss=0.1597, pruned_loss=0.04259, over 12737.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1388, pruned_loss=0.03315, over 1864692.47 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:01:37,719 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.0606, 5.5358, 5.4162, 6.0189, 5.5983, 4.9066, 5.9378, 4.9682], device='cuda:1'), covar=tensor([0.0296, 0.1020, 0.0370, 0.0376, 0.0787, 0.0355, 0.0438, 0.0507], device='cuda:1'), in_proj_covar=tensor([0.0180, 0.0277, 0.0206, 0.0203, 0.0187, 0.0162, 0.0292, 0.0172], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 17:01:37,727 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147790.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:01:41,335 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.7636, 2.7806, 4.4895, 3.4454, 4.6305, 4.3917, 4.3769, 4.0474], device='cuda:1'), covar=tensor([0.0682, 0.3108, 0.0912, 0.1600, 0.0603, 0.0903, 0.1322, 0.1332], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0305, 0.0387, 0.0296, 0.0363, 0.0320, 0.0356, 0.0292], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:01:52,236 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8211, 1.5231, 2.0598, 1.6055, 1.8863, 1.4495, 1.6817, 1.9373], device='cuda:1'), covar=tensor([0.3158, 0.2808, 0.0579, 0.2181, 0.1574, 0.1402, 0.1086, 0.1037], device='cuda:1'), in_proj_covar=tensor([0.0252, 0.0197, 0.0221, 0.0265, 0.0240, 0.0200, 0.0200, 0.0221], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 17:01:59,350 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8956, 1.5159, 3.5967, 3.3298, 3.3848, 3.6234, 3.0424, 3.6063], device='cuda:1'), covar=tensor([0.1711, 0.1834, 0.0140, 0.0300, 0.0329, 0.0157, 0.0312, 0.0149], device='cuda:1'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0170, 0.0148, 0.0142, 0.0125, 0.0124], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 17:01:59,389 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147815.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:27,883 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.007e+02 2.574e+02 3.284e+02 8.087e+02, threshold=5.147e+02, percent-clipped=4.0 2022-12-08 17:02:36,047 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147857.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:43,084 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5582, 2.1279, 2.4618, 2.6060, 2.4180, 2.0442, 2.5475, 2.2456], device='cuda:1'), covar=tensor([0.0586, 0.1316, 0.0822, 0.0665, 0.0815, 0.1560, 0.0667, 0.0959], device='cuda:1'), in_proj_covar=tensor([0.0299, 0.0266, 0.0380, 0.0337, 0.0275, 0.0314, 0.0319, 0.0284], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:02:46,226 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.3442, 5.1734, 4.7416, 5.0037, 4.9619, 5.2960, 5.3528, 5.3561], device='cuda:1'), covar=tensor([0.0768, 0.0329, 0.1935, 0.2309, 0.0761, 0.0753, 0.0646, 0.0675], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0286, 0.0457, 0.0576, 0.0363, 0.0468, 0.0402, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:02:52,275 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147876.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:54,612 INFO [train.py:873] (1/4) Epoch 20, batch 4200, loss[loss=0.1188, simple_loss=0.1495, pruned_loss=0.04404, over 6054.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.1388, pruned_loss=0.03275, over 1868830.75 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:03:10,369 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147897.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:22,143 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:28,190 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:52,201 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:55,540 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 1.943e+02 2.282e+02 3.039e+02 5.429e+02, threshold=4.564e+02, percent-clipped=1.0 2022-12-08 17:04:10,010 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147966.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:15,527 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:21,715 INFO [train.py:873] (1/4) Epoch 20, batch 4300, loss[loss=0.08595, simple_loss=0.1311, pruned_loss=0.02042, over 14278.00 frames. ], tot_loss[loss=0.103, simple_loss=0.1398, pruned_loss=0.03312, over 1923409.79 frames. ], batch size: 44, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:04:24,261 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6833, 3.5366, 3.2760, 3.3904, 3.6203, 3.6257, 3.6481, 3.6757], device='cuda:1'), covar=tensor([0.0823, 0.0559, 0.1909, 0.2416, 0.0724, 0.0901, 0.0985, 0.0836], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0285, 0.0457, 0.0574, 0.0363, 0.0469, 0.0401, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:04:32,557 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5321, 3.7379, 4.3145, 3.2972, 2.6386, 3.6670, 2.0168, 3.7553], device='cuda:1'), covar=tensor([0.0927, 0.0688, 0.0531, 0.1370, 0.1832, 0.1107, 0.2969, 0.0947], device='cuda:1'), in_proj_covar=tensor([0.0088, 0.0103, 0.0097, 0.0101, 0.0116, 0.0093, 0.0116, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 17:04:42,293 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8422, 1.7339, 1.6255, 1.8585, 1.8134, 1.8597, 1.7975, 1.6356], device='cuda:1'), covar=tensor([0.1388, 0.1210, 0.2254, 0.1255, 0.1458, 0.0972, 0.1708, 0.0997], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0285, 0.0259, 0.0291, 0.0323, 0.0306, 0.0256, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:04:52,063 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148014.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:52,939 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:05:22,415 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.071e+02 2.502e+02 3.240e+02 4.957e+02, threshold=5.004e+02, percent-clipped=3.0 2022-12-08 17:05:35,028 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:05:48,146 INFO [train.py:873] (1/4) Epoch 20, batch 4400, loss[loss=0.1184, simple_loss=0.1401, pruned_loss=0.0484, over 4952.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1391, pruned_loss=0.03199, over 2001146.91 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:05:58,147 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148090.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:00,673 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.6271, 5.4023, 5.0385, 5.6638, 5.2719, 5.0306, 5.6760, 5.3176], device='cuda:1'), covar=tensor([0.0551, 0.0746, 0.0807, 0.0473, 0.0591, 0.0429, 0.0429, 0.0634], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0150, 0.0150, 0.0166, 0.0152, 0.0129, 0.0173, 0.0153], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:06:16,514 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148111.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:17,743 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 17:06:39,823 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148138.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:40,191 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-08 17:06:49,732 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.050e+02 2.451e+02 3.050e+02 8.671e+02, threshold=4.901e+02, percent-clipped=4.0 2022-12-08 17:07:08,035 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:07:15,616 INFO [train.py:873] (1/4) Epoch 20, batch 4500, loss[loss=0.09254, simple_loss=0.1303, pruned_loss=0.02741, over 14290.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1391, pruned_loss=0.03189, over 1995655.00 frames. ], batch size: 76, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:07:45,047 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:07:47,515 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 17:08:00,541 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 17:08:17,145 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.029e+02 2.580e+02 3.112e+02 6.429e+02, threshold=5.160e+02, percent-clipped=2.0 2022-12-08 17:08:32,443 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148267.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:08:42,531 INFO [train.py:873] (1/4) Epoch 20, batch 4600, loss[loss=0.1001, simple_loss=0.1422, pruned_loss=0.02899, over 14283.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1394, pruned_loss=0.03181, over 1986811.86 frames. ], batch size: 44, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:08:49,595 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:13,738 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:35,646 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 17:09:42,763 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148348.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:44,259 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 2.179e+02 2.688e+02 3.360e+02 6.935e+02, threshold=5.375e+02, percent-clipped=4.0 2022-12-08 17:09:55,114 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:10:09,363 INFO [train.py:873] (1/4) Epoch 20, batch 4700, loss[loss=0.09437, simple_loss=0.1311, pruned_loss=0.02884, over 5025.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.139, pruned_loss=0.03211, over 1929775.26 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:10:52,736 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:07,836 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([5.9607, 5.7468, 5.3423, 5.9449, 5.4586, 5.2504, 6.0118, 5.6433], device='cuda:1'), covar=tensor([0.0506, 0.0806, 0.0720, 0.0377, 0.0590, 0.0409, 0.0449, 0.0613], device='cuda:1'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0167, 0.0153, 0.0129, 0.0174, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:11:10,543 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.097e+02 2.538e+02 3.093e+02 5.864e+02, threshold=5.077e+02, percent-clipped=2.0 2022-12-08 17:11:28,976 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:35,537 INFO [train.py:873] (1/4) Epoch 20, batch 4800, loss[loss=0.08722, simple_loss=0.1346, pruned_loss=0.01991, over 14420.00 frames. ], tot_loss[loss=0.1006, simple_loss=0.1383, pruned_loss=0.03147, over 1976838.08 frames. ], batch size: 41, lr: 3.90e-03, grad_scale: 8.0 2022-12-08 17:11:45,463 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148490.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:52,323 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148498.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:05,485 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:10,309 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148519.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:24,451 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1364, 1.8085, 3.1897, 2.4164, 3.1411, 1.8527, 2.5101, 3.1581], device='cuda:1'), covar=tensor([0.0862, 0.3723, 0.0743, 0.3768, 0.0891, 0.2722, 0.1285, 0.0794], device='cuda:1'), in_proj_covar=tensor([0.0250, 0.0197, 0.0221, 0.0263, 0.0238, 0.0198, 0.0199, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 17:12:37,156 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2022-12-08 17:12:37,249 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.096e+02 2.575e+02 3.014e+02 5.003e+02, threshold=5.149e+02, percent-clipped=0.0 2022-12-08 17:12:45,294 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148559.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:46,815 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148561.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:52,323 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148567.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:13:02,972 INFO [train.py:873] (1/4) Epoch 20, batch 4900, loss[loss=0.1224, simple_loss=0.1574, pruned_loss=0.04373, over 14289.00 frames. ], tot_loss[loss=0.1005, simple_loss=0.1385, pruned_loss=0.03128, over 2026698.22 frames. ], batch size: 25, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:13:33,594 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148615.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:13:35,387 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7815, 0.8237, 0.7223, 0.8410, 0.8476, 0.4904, 0.7842, 0.7990], device='cuda:1'), covar=tensor([0.0525, 0.0600, 0.0504, 0.0479, 0.0338, 0.0448, 0.1026, 0.0810], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0038, 0.0043, 0.0035, 0.0038, 0.0052, 0.0040, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 17:13:57,714 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:14:04,642 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.967e+02 2.434e+02 3.108e+02 5.007e+02, threshold=4.868e+02, percent-clipped=0.0 2022-12-08 17:14:04,898 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148651.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:14:11,159 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 17:14:16,755 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1491, 1.2326, 1.1336, 1.2615, 1.3263, 0.8745, 1.0388, 1.1063], device='cuda:1'), covar=tensor([0.0744, 0.0749, 0.0589, 0.0694, 0.0634, 0.0783, 0.1030, 0.0811], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0038, 0.0042, 0.0035, 0.0038, 0.0052, 0.0040, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 17:14:29,318 INFO [train.py:873] (1/4) Epoch 20, batch 5000, loss[loss=0.1145, simple_loss=0.1426, pruned_loss=0.0432, over 14156.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1388, pruned_loss=0.03171, over 1986158.14 frames. ], batch size: 99, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:14:58,843 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148712.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:02,385 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148716.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:12,140 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1989, 2.0552, 2.1328, 2.2328, 2.0988, 2.0717, 2.2732, 1.9185], device='cuda:1'), covar=tensor([0.0999, 0.1334, 0.0813, 0.0779, 0.1014, 0.0776, 0.0976, 0.0757], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0276, 0.0204, 0.0202, 0.0186, 0.0163, 0.0291, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 17:15:23,531 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2022-12-08 17:15:33,631 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.005e+02 2.005e+02 2.534e+02 3.099e+02 5.530e+02, threshold=5.067e+02, percent-clipped=2.0 2022-12-08 17:15:51,081 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7747, 1.1053, 1.2936, 1.2489, 0.8784, 1.2223, 0.9958, 0.8108], device='cuda:1'), covar=tensor([0.1948, 0.1055, 0.0434, 0.0538, 0.2354, 0.1392, 0.1825, 0.1658], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:15:56,385 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:58,225 INFO [train.py:873] (1/4) Epoch 20, batch 5100, loss[loss=0.0866, simple_loss=0.1351, pruned_loss=0.01902, over 14278.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1391, pruned_loss=0.03205, over 1958708.64 frames. ], batch size: 31, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:16:03,525 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148785.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:16:05,723 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 17:16:37,788 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 17:16:43,896 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 17:17:01,004 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.285e+02 2.852e+02 3.620e+02 6.279e+02, threshold=5.703e+02, percent-clipped=4.0 2022-12-08 17:17:03,872 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148854.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:17:25,637 INFO [train.py:873] (1/4) Epoch 20, batch 5200, loss[loss=0.1519, simple_loss=0.1641, pruned_loss=0.06978, over 8597.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1395, pruned_loss=0.0322, over 2008841.59 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 8.0 2022-12-08 17:17:56,912 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2022-12-08 17:18:21,965 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148943.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:18:24,849 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5773, 3.4484, 3.1798, 3.3423, 3.5390, 3.5760, 3.5713, 3.6158], device='cuda:1'), covar=tensor([0.0918, 0.0596, 0.1894, 0.2103, 0.0809, 0.0851, 0.0966, 0.0766], device='cuda:1'), in_proj_covar=tensor([0.0399, 0.0285, 0.0450, 0.0570, 0.0363, 0.0464, 0.0400, 0.0400], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:18:29,832 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 1.963e+02 2.482e+02 3.012e+02 7.254e+02, threshold=4.965e+02, percent-clipped=1.0 2022-12-08 17:18:53,063 INFO [train.py:873] (1/4) Epoch 20, batch 5300, loss[loss=0.09395, simple_loss=0.1341, pruned_loss=0.02688, over 14250.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.139, pruned_loss=0.03176, over 1992974.86 frames. ], batch size: 39, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:19:03,317 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148991.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:19:16,983 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:19:27,278 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 17:19:33,368 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9425, 2.5557, 2.7594, 1.9004, 2.3996, 2.6568, 2.9517, 2.4334], device='cuda:1'), covar=tensor([0.0686, 0.0774, 0.0811, 0.1223, 0.1120, 0.0779, 0.0629, 0.1165], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0168, 0.0140, 0.0126, 0.0145, 0.0158, 0.0140, 0.0142], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 17:19:57,357 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.154e+02 2.538e+02 2.985e+02 9.704e+02, threshold=5.077e+02, percent-clipped=4.0 2022-12-08 17:20:00,013 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 17:20:03,118 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6886, 3.5480, 4.2470, 3.2566, 2.7782, 3.6992, 2.1220, 3.6532], device='cuda:1'), covar=tensor([0.1282, 0.0882, 0.0531, 0.1297, 0.1518, 0.0863, 0.2542, 0.1436], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0104, 0.0098, 0.0103, 0.0116, 0.0094, 0.0117, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 17:20:04,302 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2022-12-08 17:20:14,477 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:20:20,460 INFO [train.py:873] (1/4) Epoch 20, batch 5400, loss[loss=0.1023, simple_loss=0.1404, pruned_loss=0.03207, over 14228.00 frames. ], tot_loss[loss=0.1002, simple_loss=0.1385, pruned_loss=0.03092, over 2037485.41 frames. ], batch size: 80, lr: 3.90e-03, grad_scale: 2.0 2022-12-08 17:20:25,685 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149085.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:20:58,834 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 17:20:59,967 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9366, 4.3648, 4.5892, 4.9377, 4.5225, 4.2122, 4.9047, 4.0584], device='cuda:1'), covar=tensor([0.0915, 0.1681, 0.0862, 0.0933, 0.1299, 0.0764, 0.0903, 0.0992], device='cuda:1'), in_proj_covar=tensor([0.0181, 0.0276, 0.0205, 0.0203, 0.0187, 0.0163, 0.0292, 0.0170], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 17:21:07,668 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149133.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:21:24,502 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7853, 3.3771, 3.1862, 3.0301, 2.2443, 3.2428, 3.0348, 1.5804], device='cuda:1'), covar=tensor([0.1511, 0.0905, 0.1067, 0.1579, 0.1176, 0.0719, 0.1202, 0.2191], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:21:25,134 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.118e+02 1.972e+02 2.518e+02 2.987e+02 6.531e+02, threshold=5.036e+02, percent-clipped=3.0 2022-12-08 17:21:26,158 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149154.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:21:48,144 INFO [train.py:873] (1/4) Epoch 20, batch 5500, loss[loss=0.07677, simple_loss=0.1233, pruned_loss=0.01512, over 14179.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1388, pruned_loss=0.03139, over 1978036.83 frames. ], batch size: 35, lr: 3.90e-03, grad_scale: 2.0 2022-12-08 17:22:07,506 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 17:22:07,915 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149202.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:22:39,357 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3500, 1.8973, 2.3427, 2.4583, 2.3146, 2.0942, 2.0077, 2.0812], device='cuda:1'), covar=tensor([0.0329, 0.0979, 0.0274, 0.0342, 0.0314, 0.0470, 0.0408, 0.0732], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0035], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 17:22:43,492 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2252, 2.1023, 2.1280, 2.2482, 2.1378, 2.1296, 2.3026, 1.9190], device='cuda:1'), covar=tensor([0.0931, 0.1325, 0.0866, 0.0811, 0.0996, 0.0722, 0.0906, 0.0800], device='cuda:1'), in_proj_covar=tensor([0.0182, 0.0278, 0.0205, 0.0203, 0.0188, 0.0164, 0.0294, 0.0171], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 17:22:52,116 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.034e+02 2.619e+02 3.145e+02 6.128e+02, threshold=5.238e+02, percent-clipped=4.0 2022-12-08 17:22:59,918 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2022-12-08 17:23:14,998 INFO [train.py:873] (1/4) Epoch 20, batch 5600, loss[loss=0.09004, simple_loss=0.1327, pruned_loss=0.02367, over 14335.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1389, pruned_loss=0.03168, over 1930336.55 frames. ], batch size: 73, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:23:22,083 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7072, 2.0400, 2.0680, 2.1533, 1.8844, 2.0521, 1.8063, 1.4553], device='cuda:1'), covar=tensor([0.0797, 0.0911, 0.0682, 0.0568, 0.1153, 0.0889, 0.1478, 0.1793], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0101, 0.0092, 0.0104, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:23:29,265 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9668, 1.5543, 1.9326, 1.4331, 1.6831, 2.0416, 1.8446, 1.7711], device='cuda:1'), covar=tensor([0.0901, 0.0720, 0.0748, 0.1191, 0.1729, 0.0969, 0.0813, 0.1625], device='cuda:1'), in_proj_covar=tensor([0.0156, 0.0169, 0.0141, 0.0126, 0.0146, 0.0158, 0.0140, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 17:23:32,143 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.3923, 2.6991, 2.6270, 2.7674, 2.2109, 2.7258, 2.5043, 1.5091], device='cuda:1'), covar=tensor([0.1005, 0.0968, 0.0643, 0.0509, 0.0977, 0.0668, 0.1023, 0.1881], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0101, 0.0092, 0.0104, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:23:35,711 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1267, 1.1580, 1.0514, 1.2624, 1.3127, 0.8585, 1.0182, 1.0940], device='cuda:1'), covar=tensor([0.0738, 0.0853, 0.0688, 0.0557, 0.0462, 0.0766, 0.1094, 0.0824], device='cuda:1'), in_proj_covar=tensor([0.0040, 0.0038, 0.0043, 0.0035, 0.0037, 0.0052, 0.0040, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 17:23:39,987 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:23:45,765 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-08 17:24:13,849 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.40 vs. limit=2.0 2022-12-08 17:24:19,494 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 1.989e+02 2.324e+02 2.779e+02 6.746e+02, threshold=4.648e+02, percent-clipped=1.0 2022-12-08 17:24:21,362 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149355.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:24:27,975 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 17:24:36,165 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149372.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:24:37,045 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.7455, 1.0692, 1.2421, 1.2592, 0.8390, 1.2232, 1.0141, 0.8127], device='cuda:1'), covar=tensor([0.1735, 0.1003, 0.0533, 0.0561, 0.2158, 0.1338, 0.1398, 0.1418], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:24:41,133 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([6.0658, 5.4402, 5.4348, 5.9775, 5.5102, 4.9424, 5.8880, 4.9462], device='cuda:1'), covar=tensor([0.0307, 0.0962, 0.0352, 0.0423, 0.0815, 0.0340, 0.0510, 0.0501], device='cuda:1'), in_proj_covar=tensor([0.0179, 0.0274, 0.0203, 0.0201, 0.0186, 0.0162, 0.0291, 0.0169], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:1') 2022-12-08 17:24:42,125 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1713, 2.1046, 3.1469, 3.2791, 3.1158, 2.2197, 3.1267, 2.4299], device='cuda:1'), covar=tensor([0.0564, 0.1502, 0.0861, 0.0550, 0.0760, 0.1890, 0.0551, 0.1179], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0261, 0.0375, 0.0331, 0.0271, 0.0307, 0.0314, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:24:42,616 INFO [train.py:873] (1/4) Epoch 20, batch 5700, loss[loss=0.1103, simple_loss=0.1431, pruned_loss=0.03874, over 11977.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1386, pruned_loss=0.03154, over 1976993.74 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:24:54,032 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.0600, 2.0450, 2.2003, 1.4811, 1.5610, 2.0260, 1.3301, 2.0713], device='cuda:1'), covar=tensor([0.0986, 0.1532, 0.0831, 0.2349, 0.2891, 0.1107, 0.3010, 0.0947], device='cuda:1'), in_proj_covar=tensor([0.0090, 0.0104, 0.0098, 0.0103, 0.0117, 0.0094, 0.0118, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 17:24:59,141 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5361, 1.8634, 2.5472, 2.0761, 2.4892, 2.4186, 2.3013, 2.2681], device='cuda:1'), covar=tensor([0.0906, 0.2707, 0.1054, 0.1636, 0.0682, 0.1349, 0.0785, 0.1444], device='cuda:1'), in_proj_covar=tensor([0.0350, 0.0306, 0.0388, 0.0298, 0.0363, 0.0320, 0.0359, 0.0293], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:25:18,495 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:25:29,030 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.8156, 1.6496, 4.5776, 2.0271, 4.4533, 4.9985, 4.4869, 5.2712], device='cuda:1'), covar=tensor([0.0206, 0.3317, 0.0296, 0.2239, 0.0291, 0.0278, 0.0263, 0.0131], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0158, 0.0163, 0.0170, 0.0168, 0.0182, 0.0134, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:25:34,917 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.5690, 1.5478, 2.7386, 1.5489, 2.7525, 2.7259, 2.0580, 2.8779], device='cuda:1'), covar=tensor([0.0304, 0.2668, 0.0452, 0.1846, 0.0433, 0.0538, 0.1151, 0.0282], device='cuda:1'), in_proj_covar=tensor([0.0175, 0.0158, 0.0163, 0.0170, 0.0168, 0.0182, 0.0134, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:25:47,185 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 7.719e+01 2.141e+02 2.467e+02 2.944e+02 6.062e+02, threshold=4.935e+02, percent-clipped=6.0 2022-12-08 17:26:10,319 INFO [train.py:873] (1/4) Epoch 20, batch 5800, loss[loss=0.09995, simple_loss=0.1396, pruned_loss=0.03014, over 14363.00 frames. ], tot_loss[loss=0.1005, simple_loss=0.1383, pruned_loss=0.0314, over 2033964.85 frames. ], batch size: 41, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:26:23,704 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 17:26:25,729 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 17:26:40,980 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149514.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:26:57,657 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2022-12-08 17:27:06,396 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([0.8828, 0.8075, 0.8148, 0.8250, 0.7849, 0.4506, 0.5323, 0.6696], device='cuda:1'), covar=tensor([0.0204, 0.0188, 0.0167, 0.0207, 0.0188, 0.0359, 0.0264, 0.0296], device='cuda:1'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:1') 2022-12-08 17:27:08,959 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.9357, 5.0118, 5.3641, 4.5719, 5.1721, 5.3555, 2.1976, 4.7745], device='cuda:1'), covar=tensor([0.0258, 0.0246, 0.0256, 0.0364, 0.0241, 0.0115, 0.2765, 0.0245], device='cuda:1'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0151, 0.0209, 0.0145, 0.0158, 0.0196], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 17:27:15,062 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.003e+02 2.561e+02 2.963e+02 6.960e+02, threshold=5.123e+02, percent-clipped=3.0 2022-12-08 17:27:25,719 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149565.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:27:34,505 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:27:38,141 INFO [train.py:873] (1/4) Epoch 20, batch 5900, loss[loss=0.09229, simple_loss=0.1354, pruned_loss=0.0246, over 9494.00 frames. ], tot_loss[loss=0.1006, simple_loss=0.1385, pruned_loss=0.03139, over 2022449.74 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:28:18,180 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 17:28:20,674 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149626.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:28:24,835 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.3278, 4.4463, 4.6710, 4.0937, 4.5480, 4.7965, 1.8583, 4.2076], device='cuda:1'), covar=tensor([0.0334, 0.0355, 0.0396, 0.0521, 0.0320, 0.0198, 0.3206, 0.0310], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0177, 0.0150, 0.0152, 0.0211, 0.0146, 0.0159, 0.0197], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 17:28:43,779 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.112e+02 2.459e+02 3.260e+02 5.779e+02, threshold=4.919e+02, percent-clipped=5.0 2022-12-08 17:28:50,040 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 17:29:07,357 INFO [train.py:873] (1/4) Epoch 20, batch 6000, loss[loss=0.08696, simple_loss=0.1283, pruned_loss=0.0228, over 14241.00 frames. ], tot_loss[loss=0.09994, simple_loss=0.1379, pruned_loss=0.03102, over 1985483.96 frames. ], batch size: 18, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:29:07,358 INFO [train.py:896] (1/4) Computing validation loss 2022-12-08 17:29:21,941 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.1940, 1.6914, 4.2039, 2.3996, 4.2987, 4.4282, 3.2744, 4.7524], device='cuda:1'), covar=tensor([0.0190, 0.3373, 0.0254, 0.1762, 0.0156, 0.0157, 0.0327, 0.0098], device='cuda:1'), in_proj_covar=tensor([0.0173, 0.0157, 0.0161, 0.0169, 0.0167, 0.0180, 0.0133, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:29:22,626 INFO [train.py:905] (1/4) Epoch 20, validation: loss=0.1445, simple_loss=0.1806, pruned_loss=0.0542, over 857387.00 frames. 2022-12-08 17:29:22,627 INFO [train.py:906] (1/4) Maximum memory allocated so far is 18076MB 2022-12-08 17:30:22,026 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149746.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:30:27,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.065e+02 2.623e+02 3.098e+02 9.406e+02, threshold=5.245e+02, percent-clipped=5.0 2022-12-08 17:30:50,977 INFO [train.py:873] (1/4) Epoch 20, batch 6100, loss[loss=0.122, simple_loss=0.1256, pruned_loss=0.05918, over 1217.00 frames. ], tot_loss[loss=0.09972, simple_loss=0.1377, pruned_loss=0.03086, over 1976860.64 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:31:05,584 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2022-12-08 17:31:16,086 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149807.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 17:31:18,586 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9730, 1.7491, 3.0997, 2.2760, 3.0424, 1.7942, 2.4064, 3.0261], device='cuda:1'), covar=tensor([0.1067, 0.3886, 0.0744, 0.4152, 0.1073, 0.3176, 0.1380, 0.0805], device='cuda:1'), in_proj_covar=tensor([0.0253, 0.0196, 0.0224, 0.0266, 0.0240, 0.0200, 0.0202, 0.0220], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 17:31:54,308 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2106, 3.0030, 2.7717, 2.9748, 3.1554, 3.1316, 3.1816, 3.1960], device='cuda:1'), covar=tensor([0.1027, 0.0800, 0.2278, 0.2612, 0.1006, 0.1141, 0.1284, 0.0946], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0285, 0.0452, 0.0576, 0.0364, 0.0465, 0.0399, 0.0403], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:31:56,635 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.144e+02 2.583e+02 3.219e+02 5.785e+02, threshold=5.165e+02, percent-clipped=1.0 2022-12-08 17:31:56,800 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.2021, 3.9733, 3.6611, 3.8935, 4.0985, 4.1144, 4.1791, 4.1766], device='cuda:1'), covar=tensor([0.0953, 0.0536, 0.2047, 0.2556, 0.0737, 0.0925, 0.0947, 0.0847], device='cuda:1'), in_proj_covar=tensor([0.0400, 0.0285, 0.0452, 0.0576, 0.0363, 0.0464, 0.0399, 0.0403], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:32:11,630 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:32:19,290 INFO [train.py:873] (1/4) Epoch 20, batch 6200, loss[loss=0.1047, simple_loss=0.1326, pruned_loss=0.03839, over 5959.00 frames. ], tot_loss[loss=0.09958, simple_loss=0.1376, pruned_loss=0.03079, over 1994897.53 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:32:25,574 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 17:32:49,513 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-08 17:32:57,013 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149921.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 17:33:20,150 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 17:33:24,558 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.165e+02 2.069e+02 2.540e+02 3.004e+02 8.039e+02, threshold=5.080e+02, percent-clipped=2.0 2022-12-08 17:33:47,751 INFO [train.py:873] (1/4) Epoch 20, batch 6300, loss[loss=0.0928, simple_loss=0.1277, pruned_loss=0.02894, over 6954.00 frames. ], tot_loss[loss=0.09939, simple_loss=0.1375, pruned_loss=0.03066, over 1983803.71 frames. ], batch size: 100, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:34:20,255 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8883, 1.6401, 3.4171, 3.0378, 3.2435, 3.4419, 2.8088, 3.4301], device='cuda:1'), covar=tensor([0.1736, 0.1756, 0.0145, 0.0374, 0.0361, 0.0178, 0.0347, 0.0169], device='cuda:1'), in_proj_covar=tensor([0.0145, 0.0156, 0.0131, 0.0168, 0.0149, 0.0141, 0.0124, 0.0123], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:1') 2022-12-08 17:34:31,036 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.7448, 1.4511, 2.9463, 1.6249, 2.9954, 2.9100, 2.2185, 3.1026], device='cuda:1'), covar=tensor([0.0302, 0.2889, 0.0402, 0.2005, 0.0362, 0.0469, 0.1086, 0.0240], device='cuda:1'), in_proj_covar=tensor([0.0174, 0.0158, 0.0162, 0.0170, 0.0167, 0.0181, 0.0134, 0.0154], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:34:56,193 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.076e+02 2.693e+02 3.406e+02 8.300e+02, threshold=5.385e+02, percent-clipped=3.0 2022-12-08 17:35:18,015 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.9208, 2.4916, 3.1521, 2.0614, 2.0776, 2.7706, 1.6492, 2.6849], device='cuda:1'), covar=tensor([0.0765, 0.1234, 0.0627, 0.2136, 0.2065, 0.0831, 0.3060, 0.0953], device='cuda:1'), in_proj_covar=tensor([0.0090, 0.0104, 0.0098, 0.0103, 0.0116, 0.0095, 0.0117, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 17:35:18,742 INFO [train.py:873] (1/4) Epoch 20, batch 6400, loss[loss=0.08859, simple_loss=0.1335, pruned_loss=0.02183, over 14254.00 frames. ], tot_loss[loss=0.09934, simple_loss=0.1374, pruned_loss=0.03062, over 1983866.21 frames. ], batch size: 80, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:35:18,924 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150079.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:35:39,415 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150102.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:36:12,281 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150140.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:36:23,032 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.042e+02 2.543e+02 3.172e+02 7.027e+02, threshold=5.086e+02, percent-clipped=3.0 2022-12-08 17:36:24,961 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.8432, 1.6375, 1.7986, 1.9621, 1.5219, 1.7146, 1.7380, 1.8277], device='cuda:1'), covar=tensor([0.0277, 0.0433, 0.0278, 0.0286, 0.0470, 0.0513, 0.0345, 0.0295], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0261, 0.0375, 0.0334, 0.0271, 0.0309, 0.0314, 0.0278], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:36:32,501 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.7506, 2.2297, 3.6633, 3.8834, 3.5948, 2.2941, 3.7765, 2.7878], device='cuda:1'), covar=tensor([0.0492, 0.1421, 0.0934, 0.0491, 0.0648, 0.2027, 0.0445, 0.1078], device='cuda:1'), in_proj_covar=tensor([0.0293, 0.0261, 0.0375, 0.0333, 0.0271, 0.0309, 0.0313, 0.0277], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:36:38,277 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:36:39,961 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.9423, 2.1623, 2.2779, 2.3253, 2.0264, 2.3070, 2.1097, 1.3977], device='cuda:1'), covar=tensor([0.0875, 0.0948, 0.0823, 0.0503, 0.0983, 0.0640, 0.1049, 0.1872], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0093, 0.0072, 0.0078, 0.0101, 0.0093, 0.0103, 0.0097], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:36:46,071 INFO [train.py:873] (1/4) Epoch 20, batch 6500, loss[loss=0.1044, simple_loss=0.1145, pruned_loss=0.04717, over 2670.00 frames. ], tot_loss[loss=0.1009, simple_loss=0.1385, pruned_loss=0.03167, over 1919265.45 frames. ], batch size: 100, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:37:20,281 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:37:23,022 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:37:36,777 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.1503, 3.5328, 3.2983, 3.5730, 2.5820, 3.5862, 3.3487, 1.9564], device='cuda:1'), covar=tensor([0.1263, 0.0980, 0.0940, 0.0492, 0.0924, 0.0492, 0.0799, 0.1648], device='cuda:1'), in_proj_covar=tensor([0.0134, 0.0093, 0.0073, 0.0078, 0.0101, 0.0093, 0.0103, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:37:49,882 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.2936, 1.8885, 3.4415, 2.4766, 3.3404, 1.9159, 2.6626, 3.3183], device='cuda:1'), covar=tensor([0.0712, 0.3675, 0.0583, 0.3841, 0.0698, 0.2923, 0.1315, 0.0653], device='cuda:1'), in_proj_covar=tensor([0.0249, 0.0193, 0.0222, 0.0261, 0.0238, 0.0197, 0.0200, 0.0216], device='cuda:1'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:1') 2022-12-08 17:37:50,499 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.155e+02 2.479e+02 3.022e+02 7.876e+02, threshold=4.957e+02, percent-clipped=4.0 2022-12-08 17:38:04,370 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:38:08,968 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150274.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:38:13,125 INFO [train.py:873] (1/4) Epoch 20, batch 6600, loss[loss=0.09833, simple_loss=0.1442, pruned_loss=0.02622, over 14235.00 frames. ], tot_loss[loss=0.1005, simple_loss=0.1379, pruned_loss=0.03158, over 1892052.70 frames. ], batch size: 35, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:38:31,191 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.1405, 2.0151, 2.2206, 2.3227, 1.9355, 2.0020, 2.1866, 2.1894], device='cuda:1'), covar=tensor([0.0409, 0.0744, 0.0390, 0.0428, 0.0678, 0.0955, 0.0567, 0.0451], device='cuda:1'), in_proj_covar=tensor([0.0295, 0.0262, 0.0378, 0.0335, 0.0273, 0.0310, 0.0315, 0.0279], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:39:02,451 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:39:13,413 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 17:39:18,406 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.127e+02 2.141e+02 2.452e+02 3.103e+02 5.286e+02, threshold=4.904e+02, percent-clipped=1.0 2022-12-08 17:39:22,845 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.5555, 2.3916, 3.4638, 2.6788, 3.3563, 3.3037, 3.1798, 2.8665], device='cuda:1'), covar=tensor([0.0934, 0.2557, 0.1024, 0.1683, 0.0882, 0.1081, 0.1276, 0.1657], device='cuda:1'), in_proj_covar=tensor([0.0348, 0.0306, 0.0385, 0.0297, 0.0359, 0.0319, 0.0355, 0.0290], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:39:41,069 INFO [train.py:873] (1/4) Epoch 20, batch 6700, loss[loss=0.07798, simple_loss=0.1263, pruned_loss=0.01483, over 14166.00 frames. ], tot_loss[loss=0.1002, simple_loss=0.1378, pruned_loss=0.03134, over 1897544.66 frames. ], batch size: 35, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:39:48,214 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 17:40:01,122 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150402.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:40:29,803 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:40:40,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.2755, 2.3014, 1.9507, 2.3809, 2.2148, 2.2848, 2.1820, 1.9676], device='cuda:1'), covar=tensor([0.1224, 0.1189, 0.2018, 0.1036, 0.1238, 0.0964, 0.1378, 0.1588], device='cuda:1'), in_proj_covar=tensor([0.0282, 0.0285, 0.0257, 0.0290, 0.0322, 0.0301, 0.0257, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:40:42,817 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150450.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:40:45,137 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.053e+02 2.408e+02 2.928e+02 1.041e+03, threshold=4.817e+02, percent-clipped=2.0 2022-12-08 17:41:07,584 INFO [train.py:873] (1/4) Epoch 20, batch 6800, loss[loss=0.105, simple_loss=0.1357, pruned_loss=0.03712, over 4942.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1378, pruned_loss=0.03125, over 1937713.50 frames. ], batch size: 100, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:41:36,624 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2022-12-08 17:41:47,773 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.8057, 2.2942, 3.7732, 3.9768, 3.6231, 2.2437, 3.8866, 2.8039], device='cuda:1'), covar=tensor([0.0554, 0.1513, 0.0996, 0.0560, 0.0764, 0.2319, 0.0444, 0.1276], device='cuda:1'), in_proj_covar=tensor([0.0296, 0.0263, 0.0380, 0.0336, 0.0274, 0.0311, 0.0316, 0.0280], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:41:57,931 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150537.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:42:10,909 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2022-12-08 17:42:12,046 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.195e+02 2.476e+02 3.043e+02 9.162e+02, threshold=4.952e+02, percent-clipped=3.0 2022-12-08 17:42:18,444 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 17:42:23,331 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6938, 2.4119, 2.6249, 1.7841, 2.2763, 2.5652, 2.6812, 2.3319], device='cuda:1'), covar=tensor([0.0818, 0.0671, 0.0862, 0.1298, 0.1240, 0.0846, 0.0723, 0.1224], device='cuda:1'), in_proj_covar=tensor([0.0155, 0.0168, 0.0141, 0.0125, 0.0145, 0.0157, 0.0140, 0.0143], device='cuda:1'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:1') 2022-12-08 17:42:34,233 INFO [train.py:873] (1/4) Epoch 20, batch 6900, loss[loss=0.104, simple_loss=0.1458, pruned_loss=0.03107, over 14347.00 frames. ], tot_loss[loss=0.1002, simple_loss=0.1381, pruned_loss=0.03115, over 2024740.23 frames. ], batch size: 55, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:42:51,135 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150598.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:43:19,293 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:43:40,041 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 1.852e+02 2.409e+02 3.250e+02 6.967e+02, threshold=4.817e+02, percent-clipped=2.0 2022-12-08 17:43:42,755 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6495, 2.7219, 2.8040, 2.7130, 2.7764, 2.5228, 1.6301, 2.5411], device='cuda:1'), covar=tensor([0.0544, 0.0490, 0.0401, 0.0461, 0.0369, 0.1010, 0.2386, 0.0420], device='cuda:1'), in_proj_covar=tensor([0.0178, 0.0177, 0.0149, 0.0153, 0.0211, 0.0145, 0.0159, 0.0198], device='cuda:1'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:1') 2022-12-08 17:44:02,000 INFO [train.py:873] (1/4) Epoch 20, batch 7000, loss[loss=0.1044, simple_loss=0.1449, pruned_loss=0.03195, over 14355.00 frames. ], tot_loss[loss=0.09961, simple_loss=0.1377, pruned_loss=0.03077, over 1983861.85 frames. ], batch size: 73, lr: 3.88e-03, grad_scale: 4.0 2022-12-08 17:44:04,740 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.3594, 3.1424, 3.9998, 2.9199, 2.5714, 3.4135, 1.9485, 3.6245], device='cuda:1'), covar=tensor([0.1202, 0.0900, 0.0406, 0.1748, 0.1581, 0.0816, 0.2425, 0.1131], device='cuda:1'), in_proj_covar=tensor([0.0089, 0.0103, 0.0097, 0.0102, 0.0115, 0.0093, 0.0115, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:1') 2022-12-08 17:44:48,576 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:44:50,914 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150735.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:45:06,716 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.298e+02 2.589e+02 3.066e+02 6.150e+02, threshold=5.179e+02, percent-clipped=4.0 2022-12-08 17:45:12,926 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.5105, 3.9103, 3.0863, 4.8200, 4.2798, 4.5208, 3.9888, 3.2239], device='cuda:1'), covar=tensor([0.0600, 0.1078, 0.2917, 0.0411, 0.0652, 0.1173, 0.1050, 0.2700], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0287, 0.0258, 0.0290, 0.0324, 0.0302, 0.0258, 0.0241], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:45:28,337 INFO [train.py:873] (1/4) Epoch 20, batch 7100, loss[loss=0.1275, simple_loss=0.1174, pruned_loss=0.06879, over 1300.00 frames. ], tot_loss[loss=0.1002, simple_loss=0.1381, pruned_loss=0.03118, over 2050172.32 frames. ], batch size: 100, lr: 3.87e-03, grad_scale: 4.0 2022-12-08 17:45:31,475 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150783.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:45:34,612 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([2.6660, 3.0345, 2.8811, 2.9281, 2.3805, 3.0727, 2.9061, 1.6429], device='cuda:1'), covar=tensor([0.0938, 0.0478, 0.0821, 0.0568, 0.0794, 0.0408, 0.0682, 0.1532], device='cuda:1'), in_proj_covar=tensor([0.0133, 0.0093, 0.0073, 0.0078, 0.0101, 0.0093, 0.0103, 0.0098], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:45:37,009 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.7673, 1.7788, 1.6302, 1.9150, 1.7181, 1.7782, 1.8288, 1.6275], device='cuda:1'), covar=tensor([0.1249, 0.0947, 0.1991, 0.0985, 0.1158, 0.0865, 0.1663, 0.1093], device='cuda:1'), in_proj_covar=tensor([0.0284, 0.0287, 0.0259, 0.0291, 0.0324, 0.0302, 0.0258, 0.0242], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:45:37,436 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 17:45:40,408 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150793.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 17:45:42,957 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.6340, 3.4044, 3.1873, 3.3513, 3.5611, 3.5351, 3.5754, 3.6121], device='cuda:1'), covar=tensor([0.0837, 0.0608, 0.1854, 0.2267, 0.0780, 0.0931, 0.1061, 0.0780], device='cuda:1'), in_proj_covar=tensor([0.0399, 0.0284, 0.0453, 0.0573, 0.0361, 0.0462, 0.0401, 0.0402], device='cuda:1'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:1') 2022-12-08 17:46:33,853 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.271e+02 2.646e+02 3.296e+02 2.153e+03, threshold=5.293e+02, percent-clipped=6.0 2022-12-08 17:46:55,669 INFO [train.py:873] (1/4) Epoch 20, batch 7200, loss[loss=0.1408, simple_loss=0.1383, pruned_loss=0.07162, over 2576.00 frames. ], tot_loss[loss=0.101, simple_loss=0.1386, pruned_loss=0.03166, over 1997470.96 frames. ], batch size: 100, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:47:07,850 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150893.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:47:14,136 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150900.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:47:28,531 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 17:47:40,809 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150930.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:47:59,056 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2022-12-08 17:48:01,352 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 1.977e+02 2.439e+02 3.091e+02 5.350e+02, threshold=4.878e+02, percent-clipped=1.0 2022-12-08 17:48:07,527 INFO [zipformer.py:626] (1/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150961.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:48:17,473 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 17:48:22,756 INFO [zipformer.py:626] (1/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:48:23,465 INFO [train.py:873] (1/4) Epoch 20, batch 7300, loss[loss=0.085, simple_loss=0.1282, pruned_loss=0.02089, over 14292.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1377, pruned_loss=0.03121, over 1974782.69 frames. ], batch size: 76, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:49:29,351 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.139e+02 2.523e+02 3.177e+02 9.954e+02, threshold=5.046e+02, percent-clipped=7.0 2022-12-08 17:49:50,346 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([3.4738, 4.0547, 3.5535, 3.7592, 2.7425, 3.8207, 3.7385, 2.0235], device='cuda:1'), covar=tensor([0.1365, 0.0594, 0.0968, 0.0592, 0.0865, 0.0465, 0.0846, 0.1700], device='cuda:1'), in_proj_covar=tensor([0.0135, 0.0094, 0.0074, 0.0079, 0.0103, 0.0095, 0.0105, 0.0099], device='cuda:1'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:1') 2022-12-08 17:49:51,031 INFO [train.py:873] (1/4) Epoch 20, batch 7400, loss[loss=0.08365, simple_loss=0.1291, pruned_loss=0.0191, over 14039.00 frames. ], tot_loss[loss=0.09989, simple_loss=0.1375, pruned_loss=0.03115, over 1972193.12 frames. ], batch size: 26, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:49:59,291 INFO [zipformer.py:626] (1/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151088.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:50:16,701 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=2.51 vs. limit=5.0 2022-12-08 17:50:33,732 INFO [scaling.py:679] (1/4) Whitening: num_groups=1, num_channels=384, metric=3.46 vs. limit=5.0 2022-12-08 17:50:55,891 INFO [optim.py:369] (1/4) Clipping_scale=2.0, grad-norm quartiles 1.036e+02 2.135e+02 2.491e+02 3.146e+02 1.107e+03, threshold=4.982e+02, percent-clipped=6.0 2022-12-08 17:50:56,974 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([1.1575, 1.1335, 1.0915, 1.1730, 1.2713, 0.8861, 1.0603, 1.0117], device='cuda:1'), covar=tensor([0.0676, 0.0763, 0.0615, 0.0544, 0.0619, 0.0712, 0.0922, 0.0761], device='cuda:1'), in_proj_covar=tensor([0.0039, 0.0038, 0.0043, 0.0035, 0.0038, 0.0052, 0.0040, 0.0042], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:1') 2022-12-08 17:51:17,606 INFO [train.py:873] (1/4) Epoch 20, batch 7500, loss[loss=0.1191, simple_loss=0.1234, pruned_loss=0.05739, over 1165.00 frames. ], tot_loss[loss=0.09965, simple_loss=0.1374, pruned_loss=0.03097, over 2003682.32 frames. ], batch size: 100, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:51:21,864 INFO [zipformer.py:626] (1/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151184.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:51:29,994 INFO [zipformer.py:626] (1/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:51:30,800 INFO [zipformer.py:1414] (1/4) attn_weights_entropy = tensor([4.4438, 4.2253, 4.0977, 4.4130, 4.0696, 3.7641, 4.4830, 4.2498], device='cuda:1'), covar=tensor([0.0531, 0.0795, 0.0839, 0.0586, 0.0822, 0.0744, 0.0550, 0.0753], device='cuda:1'), in_proj_covar=tensor([0.0147, 0.0150, 0.0150, 0.0167, 0.0151, 0.0128, 0.0174, 0.0155], device='cuda:1'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:1') 2022-12-08 17:51:32,834 INFO [scaling.py:679] (1/4) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2022-12-08 17:52:04,557 INFO [train.py:1091] (1/4) Done!