clip-wiki-pretrain-10240-6000 / trainer_state.json
gowitheflow1998
upload checkpoint
73e49b9
raw
history blame contribute delete
No virus
36.4 kB
{
"best_metric": 0.7258342522680835,
"best_model_checkpoint": "./checkpoints/clip-stage1-10240/checkpoint-100",
"epoch": 0.9970089730807578,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2e-06,
"loss": 3.6467,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 4e-06,
"loss": 2.3745,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 6e-06,
"loss": 1.9228,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 8e-06,
"loss": 1.6728,
"step": 80
},
{
"epoch": 0.02,
"learning_rate": 9.999999999999999e-06,
"loss": 1.5235,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 1.2e-05,
"loss": 1.4338,
"step": 120
},
{
"epoch": 0.02,
"learning_rate": 1.4e-05,
"loss": 1.3564,
"step": 140
},
{
"epoch": 0.03,
"learning_rate": 1.6e-05,
"loss": 1.2923,
"step": 160
},
{
"epoch": 0.03,
"learning_rate": 1.8e-05,
"loss": 1.2428,
"step": 180
},
{
"epoch": 0.03,
"learning_rate": 1.9999999999999998e-05,
"loss": 1.2069,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 2.2e-05,
"loss": 1.1654,
"step": 220
},
{
"epoch": 0.04,
"learning_rate": 2.4e-05,
"loss": 1.129,
"step": 240
},
{
"epoch": 0.04,
"learning_rate": 2.6000000000000002e-05,
"loss": 1.095,
"step": 260
},
{
"epoch": 0.05,
"learning_rate": 2.8e-05,
"loss": 1.0813,
"step": 280
},
{
"epoch": 0.05,
"learning_rate": 3e-05,
"loss": 1.0522,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 2.9894736842105264e-05,
"loss": 1.0212,
"step": 320
},
{
"epoch": 0.06,
"learning_rate": 2.9789473684210527e-05,
"loss": 0.9983,
"step": 340
},
{
"epoch": 0.06,
"learning_rate": 2.968421052631579e-05,
"loss": 0.9816,
"step": 360
},
{
"epoch": 0.06,
"learning_rate": 2.957894736842105e-05,
"loss": 0.9694,
"step": 380
},
{
"epoch": 0.07,
"learning_rate": 2.9473684210526314e-05,
"loss": 0.9517,
"step": 400
},
{
"epoch": 0.07,
"learning_rate": 2.936842105263158e-05,
"loss": 0.9385,
"step": 420
},
{
"epoch": 0.07,
"learning_rate": 2.9263157894736844e-05,
"loss": 0.9146,
"step": 440
},
{
"epoch": 0.08,
"learning_rate": 2.9157894736842107e-05,
"loss": 0.9091,
"step": 460
},
{
"epoch": 0.08,
"learning_rate": 2.905263157894737e-05,
"loss": 0.8921,
"step": 480
},
{
"epoch": 0.08,
"learning_rate": 2.8947368421052634e-05,
"loss": 0.8871,
"step": 500
},
{
"epoch": 0.09,
"learning_rate": 2.8842105263157897e-05,
"loss": 0.8704,
"step": 520
},
{
"epoch": 0.09,
"learning_rate": 2.8736842105263157e-05,
"loss": 0.8601,
"step": 540
},
{
"epoch": 0.09,
"learning_rate": 2.863157894736842e-05,
"loss": 0.8561,
"step": 560
},
{
"epoch": 0.1,
"learning_rate": 2.8526315789473683e-05,
"loss": 0.8481,
"step": 580
},
{
"epoch": 0.1,
"learning_rate": 2.8421052631578946e-05,
"loss": 0.8323,
"step": 600
},
{
"epoch": 0.1,
"learning_rate": 2.8315789473684213e-05,
"loss": 0.8285,
"step": 620
},
{
"epoch": 0.11,
"learning_rate": 2.8210526315789476e-05,
"loss": 0.8279,
"step": 640
},
{
"epoch": 0.11,
"learning_rate": 2.810526315789474e-05,
"loss": 0.8166,
"step": 660
},
{
"epoch": 0.11,
"learning_rate": 2.8e-05,
"loss": 0.8072,
"step": 680
},
{
"epoch": 0.12,
"learning_rate": 2.7894736842105263e-05,
"loss": 0.7981,
"step": 700
},
{
"epoch": 0.12,
"learning_rate": 2.7789473684210526e-05,
"loss": 0.8,
"step": 720
},
{
"epoch": 0.12,
"learning_rate": 2.768421052631579e-05,
"loss": 0.7873,
"step": 740
},
{
"epoch": 0.13,
"learning_rate": 2.7578947368421053e-05,
"loss": 0.7805,
"step": 760
},
{
"epoch": 0.13,
"learning_rate": 2.7473684210526316e-05,
"loss": 0.7772,
"step": 780
},
{
"epoch": 0.13,
"learning_rate": 2.736842105263158e-05,
"loss": 0.7683,
"step": 800
},
{
"epoch": 0.14,
"learning_rate": 2.7263157894736846e-05,
"loss": 0.7711,
"step": 820
},
{
"epoch": 0.14,
"learning_rate": 2.7157894736842106e-05,
"loss": 0.7627,
"step": 840
},
{
"epoch": 0.14,
"learning_rate": 2.705263157894737e-05,
"loss": 0.7543,
"step": 860
},
{
"epoch": 0.15,
"learning_rate": 2.6947368421052632e-05,
"loss": 0.7538,
"step": 880
},
{
"epoch": 0.15,
"learning_rate": 2.6842105263157896e-05,
"loss": 0.7468,
"step": 900
},
{
"epoch": 0.15,
"learning_rate": 2.673684210526316e-05,
"loss": 0.7376,
"step": 920
},
{
"epoch": 0.16,
"learning_rate": 2.6631578947368422e-05,
"loss": 0.7427,
"step": 940
},
{
"epoch": 0.16,
"learning_rate": 2.6526315789473685e-05,
"loss": 0.7305,
"step": 960
},
{
"epoch": 0.16,
"learning_rate": 2.6421052631578945e-05,
"loss": 0.727,
"step": 980
},
{
"epoch": 0.17,
"learning_rate": 2.631578947368421e-05,
"loss": 0.7305,
"step": 1000
},
{
"epoch": 0.17,
"learning_rate": 2.6210526315789475e-05,
"loss": 0.7243,
"step": 1020
},
{
"epoch": 0.17,
"learning_rate": 2.610526315789474e-05,
"loss": 0.7189,
"step": 1040
},
{
"epoch": 0.18,
"learning_rate": 2.6000000000000002e-05,
"loss": 0.721,
"step": 1060
},
{
"epoch": 0.18,
"learning_rate": 2.5894736842105265e-05,
"loss": 0.7164,
"step": 1080
},
{
"epoch": 0.18,
"learning_rate": 2.578947368421053e-05,
"loss": 0.7171,
"step": 1100
},
{
"epoch": 0.19,
"learning_rate": 2.568421052631579e-05,
"loss": 0.7082,
"step": 1120
},
{
"epoch": 0.19,
"learning_rate": 2.557894736842105e-05,
"loss": 0.6998,
"step": 1140
},
{
"epoch": 0.19,
"learning_rate": 2.5473684210526315e-05,
"loss": 0.7025,
"step": 1160
},
{
"epoch": 0.2,
"learning_rate": 2.5368421052631578e-05,
"loss": 0.6959,
"step": 1180
},
{
"epoch": 0.2,
"learning_rate": 2.526315789473684e-05,
"loss": 0.6996,
"step": 1200
},
{
"epoch": 0.2,
"learning_rate": 2.5157894736842108e-05,
"loss": 0.6923,
"step": 1220
},
{
"epoch": 0.21,
"learning_rate": 2.505263157894737e-05,
"loss": 0.6906,
"step": 1240
},
{
"epoch": 0.21,
"learning_rate": 2.4947368421052635e-05,
"loss": 0.6894,
"step": 1260
},
{
"epoch": 0.21,
"learning_rate": 2.4842105263157894e-05,
"loss": 0.6838,
"step": 1280
},
{
"epoch": 0.22,
"learning_rate": 2.4736842105263158e-05,
"loss": 0.6803,
"step": 1300
},
{
"epoch": 0.22,
"learning_rate": 2.463157894736842e-05,
"loss": 0.68,
"step": 1320
},
{
"epoch": 0.22,
"learning_rate": 2.4526315789473684e-05,
"loss": 0.6734,
"step": 1340
},
{
"epoch": 0.23,
"learning_rate": 2.4421052631578948e-05,
"loss": 0.6732,
"step": 1360
},
{
"epoch": 0.23,
"learning_rate": 2.431578947368421e-05,
"loss": 0.673,
"step": 1380
},
{
"epoch": 0.23,
"learning_rate": 2.4210526315789474e-05,
"loss": 0.6675,
"step": 1400
},
{
"epoch": 0.24,
"learning_rate": 2.410526315789474e-05,
"loss": 0.6689,
"step": 1420
},
{
"epoch": 0.24,
"learning_rate": 2.4e-05,
"loss": 0.6671,
"step": 1440
},
{
"epoch": 0.24,
"learning_rate": 2.3894736842105264e-05,
"loss": 0.6666,
"step": 1460
},
{
"epoch": 0.25,
"learning_rate": 2.3789473684210527e-05,
"loss": 0.6582,
"step": 1480
},
{
"epoch": 0.25,
"learning_rate": 2.368421052631579e-05,
"loss": 0.6541,
"step": 1500
},
{
"epoch": 0.25,
"learning_rate": 2.3578947368421054e-05,
"loss": 0.6621,
"step": 1520
},
{
"epoch": 0.26,
"learning_rate": 2.3473684210526317e-05,
"loss": 0.6549,
"step": 1540
},
{
"epoch": 0.26,
"learning_rate": 2.336842105263158e-05,
"loss": 0.6503,
"step": 1560
},
{
"epoch": 0.26,
"learning_rate": 2.326315789473684e-05,
"loss": 0.6492,
"step": 1580
},
{
"epoch": 0.27,
"learning_rate": 2.3157894736842103e-05,
"loss": 0.6498,
"step": 1600
},
{
"epoch": 0.27,
"learning_rate": 2.3052631578947367e-05,
"loss": 0.6523,
"step": 1620
},
{
"epoch": 0.27,
"learning_rate": 2.2947368421052633e-05,
"loss": 0.6443,
"step": 1640
},
{
"epoch": 0.28,
"learning_rate": 2.2842105263157897e-05,
"loss": 0.6442,
"step": 1660
},
{
"epoch": 0.28,
"learning_rate": 2.273684210526316e-05,
"loss": 0.6459,
"step": 1680
},
{
"epoch": 0.28,
"learning_rate": 2.2631578947368423e-05,
"loss": 0.6426,
"step": 1700
},
{
"epoch": 0.29,
"learning_rate": 2.2526315789473686e-05,
"loss": 0.636,
"step": 1720
},
{
"epoch": 0.29,
"learning_rate": 2.2421052631578946e-05,
"loss": 0.6365,
"step": 1740
},
{
"epoch": 0.29,
"learning_rate": 2.231578947368421e-05,
"loss": 0.636,
"step": 1760
},
{
"epoch": 0.3,
"learning_rate": 2.2210526315789473e-05,
"loss": 0.6381,
"step": 1780
},
{
"epoch": 0.3,
"learning_rate": 2.2105263157894736e-05,
"loss": 0.6346,
"step": 1800
},
{
"epoch": 0.3,
"learning_rate": 2.2e-05,
"loss": 0.6291,
"step": 1820
},
{
"epoch": 0.31,
"learning_rate": 2.1894736842105266e-05,
"loss": 0.63,
"step": 1840
},
{
"epoch": 0.31,
"learning_rate": 2.178947368421053e-05,
"loss": 0.6296,
"step": 1860
},
{
"epoch": 0.31,
"learning_rate": 2.168421052631579e-05,
"loss": 0.6278,
"step": 1880
},
{
"epoch": 0.32,
"learning_rate": 2.1578947368421053e-05,
"loss": 0.6261,
"step": 1900
},
{
"epoch": 0.32,
"learning_rate": 2.1473684210526316e-05,
"loss": 0.6234,
"step": 1920
},
{
"epoch": 0.32,
"learning_rate": 2.136842105263158e-05,
"loss": 0.6201,
"step": 1940
},
{
"epoch": 0.33,
"learning_rate": 2.1263157894736842e-05,
"loss": 0.6187,
"step": 1960
},
{
"epoch": 0.33,
"learning_rate": 2.1157894736842106e-05,
"loss": 0.6201,
"step": 1980
},
{
"epoch": 0.33,
"learning_rate": 2.105263157894737e-05,
"loss": 0.6165,
"step": 2000
},
{
"epoch": 0.34,
"learning_rate": 2.0947368421052632e-05,
"loss": 0.6181,
"step": 2020
},
{
"epoch": 0.34,
"learning_rate": 2.0842105263157895e-05,
"loss": 0.6129,
"step": 2040
},
{
"epoch": 0.34,
"learning_rate": 2.073684210526316e-05,
"loss": 0.6158,
"step": 2060
},
{
"epoch": 0.35,
"learning_rate": 2.0631578947368422e-05,
"loss": 0.6141,
"step": 2080
},
{
"epoch": 0.35,
"learning_rate": 2.0526315789473685e-05,
"loss": 0.608,
"step": 2100
},
{
"epoch": 0.35,
"learning_rate": 2.042105263157895e-05,
"loss": 0.6077,
"step": 2120
},
{
"epoch": 0.36,
"learning_rate": 2.0315789473684212e-05,
"loss": 0.6087,
"step": 2140
},
{
"epoch": 0.36,
"learning_rate": 2.0210526315789475e-05,
"loss": 0.6066,
"step": 2160
},
{
"epoch": 0.36,
"learning_rate": 2.0105263157894735e-05,
"loss": 0.6035,
"step": 2180
},
{
"epoch": 0.37,
"learning_rate": 1.9999999999999998e-05,
"loss": 0.6006,
"step": 2200
},
{
"epoch": 0.37,
"learning_rate": 1.989473684210526e-05,
"loss": 0.6061,
"step": 2220
},
{
"epoch": 0.37,
"learning_rate": 1.9789473684210528e-05,
"loss": 0.6025,
"step": 2240
},
{
"epoch": 0.38,
"learning_rate": 1.968421052631579e-05,
"loss": 0.6021,
"step": 2260
},
{
"epoch": 0.38,
"learning_rate": 1.9578947368421055e-05,
"loss": 0.5989,
"step": 2280
},
{
"epoch": 0.38,
"learning_rate": 1.9473684210526318e-05,
"loss": 0.5998,
"step": 2300
},
{
"epoch": 0.39,
"learning_rate": 1.936842105263158e-05,
"loss": 0.597,
"step": 2320
},
{
"epoch": 0.39,
"learning_rate": 1.926315789473684e-05,
"loss": 0.5995,
"step": 2340
},
{
"epoch": 0.39,
"learning_rate": 1.9157894736842104e-05,
"loss": 0.5965,
"step": 2360
},
{
"epoch": 0.4,
"learning_rate": 1.9052631578947368e-05,
"loss": 0.5922,
"step": 2380
},
{
"epoch": 0.4,
"learning_rate": 1.894736842105263e-05,
"loss": 0.5922,
"step": 2400
},
{
"epoch": 0.4,
"learning_rate": 1.8842105263157894e-05,
"loss": 0.5964,
"step": 2420
},
{
"epoch": 0.41,
"learning_rate": 1.873684210526316e-05,
"loss": 0.5898,
"step": 2440
},
{
"epoch": 0.41,
"learning_rate": 1.8631578947368424e-05,
"loss": 0.5897,
"step": 2460
},
{
"epoch": 0.41,
"learning_rate": 1.8526315789473684e-05,
"loss": 0.5818,
"step": 2480
},
{
"epoch": 0.42,
"learning_rate": 1.8421052631578947e-05,
"loss": 0.5876,
"step": 2500
},
{
"epoch": 0.42,
"learning_rate": 1.831578947368421e-05,
"loss": 0.5868,
"step": 2520
},
{
"epoch": 0.42,
"learning_rate": 1.8210526315789474e-05,
"loss": 0.5864,
"step": 2540
},
{
"epoch": 0.43,
"learning_rate": 1.8105263157894737e-05,
"loss": 0.5851,
"step": 2560
},
{
"epoch": 0.43,
"learning_rate": 1.8e-05,
"loss": 0.5835,
"step": 2580
},
{
"epoch": 0.43,
"learning_rate": 1.7894736842105264e-05,
"loss": 0.5777,
"step": 2600
},
{
"epoch": 0.44,
"learning_rate": 1.7789473684210527e-05,
"loss": 0.5832,
"step": 2620
},
{
"epoch": 0.44,
"learning_rate": 1.7684210526315787e-05,
"loss": 0.579,
"step": 2640
},
{
"epoch": 0.44,
"learning_rate": 1.7578947368421054e-05,
"loss": 0.5833,
"step": 2660
},
{
"epoch": 0.45,
"learning_rate": 1.7473684210526317e-05,
"loss": 0.585,
"step": 2680
},
{
"epoch": 0.45,
"learning_rate": 1.736842105263158e-05,
"loss": 0.5819,
"step": 2700
},
{
"epoch": 0.45,
"learning_rate": 1.7263157894736843e-05,
"loss": 0.5744,
"step": 2720
},
{
"epoch": 0.46,
"learning_rate": 1.7157894736842107e-05,
"loss": 0.5743,
"step": 2740
},
{
"epoch": 0.46,
"learning_rate": 1.705263157894737e-05,
"loss": 0.5787,
"step": 2760
},
{
"epoch": 0.46,
"learning_rate": 1.694736842105263e-05,
"loss": 0.5737,
"step": 2780
},
{
"epoch": 0.47,
"learning_rate": 1.6842105263157893e-05,
"loss": 0.5733,
"step": 2800
},
{
"epoch": 0.47,
"learning_rate": 1.6736842105263156e-05,
"loss": 0.5711,
"step": 2820
},
{
"epoch": 0.47,
"learning_rate": 1.663157894736842e-05,
"loss": 0.5667,
"step": 2840
},
{
"epoch": 0.48,
"learning_rate": 1.6526315789473686e-05,
"loss": 0.5659,
"step": 2860
},
{
"epoch": 0.48,
"learning_rate": 1.642105263157895e-05,
"loss": 0.5685,
"step": 2880
},
{
"epoch": 0.48,
"learning_rate": 1.6315789473684213e-05,
"loss": 0.5664,
"step": 2900
},
{
"epoch": 0.49,
"learning_rate": 1.6210526315789476e-05,
"loss": 0.5691,
"step": 2920
},
{
"epoch": 0.49,
"learning_rate": 1.6105263157894736e-05,
"loss": 0.5645,
"step": 2940
},
{
"epoch": 0.49,
"learning_rate": 1.6e-05,
"loss": 0.568,
"step": 2960
},
{
"epoch": 0.5,
"learning_rate": 1.5894736842105263e-05,
"loss": 0.5614,
"step": 2980
},
{
"epoch": 0.5,
"learning_rate": 1.5789473684210526e-05,
"loss": 0.5648,
"step": 3000
},
{
"epoch": 0.5,
"learning_rate": 1.568421052631579e-05,
"loss": 0.5635,
"step": 3020
},
{
"epoch": 0.51,
"learning_rate": 1.5578947368421052e-05,
"loss": 0.5635,
"step": 3040
},
{
"epoch": 0.51,
"learning_rate": 1.547368421052632e-05,
"loss": 0.5648,
"step": 3060
},
{
"epoch": 0.51,
"learning_rate": 1.536842105263158e-05,
"loss": 0.5636,
"step": 3080
},
{
"epoch": 0.52,
"learning_rate": 1.5263157894736842e-05,
"loss": 0.5593,
"step": 3100
},
{
"epoch": 0.52,
"learning_rate": 1.5157894736842105e-05,
"loss": 0.5553,
"step": 3120
},
{
"epoch": 0.52,
"learning_rate": 1.5052631578947369e-05,
"loss": 0.5602,
"step": 3140
},
{
"epoch": 0.53,
"learning_rate": 1.4947368421052632e-05,
"loss": 0.5574,
"step": 3160
},
{
"epoch": 0.53,
"learning_rate": 1.4842105263157895e-05,
"loss": 0.558,
"step": 3180
},
{
"epoch": 0.53,
"learning_rate": 1.4736842105263157e-05,
"loss": 0.555,
"step": 3200
},
{
"epoch": 0.54,
"learning_rate": 1.4631578947368422e-05,
"loss": 0.5619,
"step": 3220
},
{
"epoch": 0.54,
"learning_rate": 1.4526315789473685e-05,
"loss": 0.5531,
"step": 3240
},
{
"epoch": 0.54,
"learning_rate": 1.4421052631578948e-05,
"loss": 0.5547,
"step": 3260
},
{
"epoch": 0.55,
"learning_rate": 1.431578947368421e-05,
"loss": 0.5551,
"step": 3280
},
{
"epoch": 0.55,
"learning_rate": 1.4210526315789473e-05,
"loss": 0.5541,
"step": 3300
},
{
"epoch": 0.55,
"learning_rate": 1.4105263157894738e-05,
"loss": 0.5512,
"step": 3320
},
{
"epoch": 0.56,
"learning_rate": 1.4e-05,
"loss": 0.5531,
"step": 3340
},
{
"epoch": 0.56,
"learning_rate": 1.3894736842105263e-05,
"loss": 0.5504,
"step": 3360
},
{
"epoch": 0.56,
"learning_rate": 1.3789473684210526e-05,
"loss": 0.5508,
"step": 3380
},
{
"epoch": 0.56,
"learning_rate": 1.368421052631579e-05,
"loss": 0.5532,
"step": 3400
},
{
"epoch": 0.57,
"learning_rate": 1.3578947368421053e-05,
"loss": 0.5458,
"step": 3420
},
{
"epoch": 0.57,
"learning_rate": 1.3473684210526316e-05,
"loss": 0.5505,
"step": 3440
},
{
"epoch": 0.57,
"learning_rate": 1.336842105263158e-05,
"loss": 0.5519,
"step": 3460
},
{
"epoch": 0.58,
"learning_rate": 1.3263157894736843e-05,
"loss": 0.5464,
"step": 3480
},
{
"epoch": 0.58,
"learning_rate": 1.3157894736842104e-05,
"loss": 0.5488,
"step": 3500
},
{
"epoch": 0.58,
"learning_rate": 1.305263157894737e-05,
"loss": 0.5475,
"step": 3520
},
{
"epoch": 0.59,
"learning_rate": 1.2947368421052633e-05,
"loss": 0.548,
"step": 3540
},
{
"epoch": 0.59,
"learning_rate": 1.2842105263157896e-05,
"loss": 0.5449,
"step": 3560
},
{
"epoch": 0.59,
"learning_rate": 1.2736842105263157e-05,
"loss": 0.5447,
"step": 3580
},
{
"epoch": 0.6,
"learning_rate": 1.263157894736842e-05,
"loss": 0.5421,
"step": 3600
},
{
"epoch": 0.6,
"learning_rate": 1.2526315789473686e-05,
"loss": 0.5487,
"step": 3620
},
{
"epoch": 0.6,
"learning_rate": 1.2421052631578947e-05,
"loss": 0.5441,
"step": 3640
},
{
"epoch": 0.61,
"learning_rate": 1.231578947368421e-05,
"loss": 0.5404,
"step": 3660
},
{
"epoch": 0.61,
"learning_rate": 1.2210526315789474e-05,
"loss": 0.5386,
"step": 3680
},
{
"epoch": 0.61,
"learning_rate": 1.2105263157894737e-05,
"loss": 0.5452,
"step": 3700
},
{
"epoch": 0.62,
"learning_rate": 1.2e-05,
"loss": 0.5417,
"step": 3720
},
{
"epoch": 0.62,
"learning_rate": 1.1894736842105264e-05,
"loss": 0.5419,
"step": 3740
},
{
"epoch": 0.62,
"learning_rate": 1.1789473684210527e-05,
"loss": 0.5409,
"step": 3760
},
{
"epoch": 0.63,
"learning_rate": 1.168421052631579e-05,
"loss": 0.5341,
"step": 3780
},
{
"epoch": 0.63,
"learning_rate": 1.1578947368421052e-05,
"loss": 0.5392,
"step": 3800
},
{
"epoch": 0.63,
"learning_rate": 1.1473684210526317e-05,
"loss": 0.5393,
"step": 3820
},
{
"epoch": 0.64,
"learning_rate": 1.136842105263158e-05,
"loss": 0.5363,
"step": 3840
},
{
"epoch": 0.64,
"learning_rate": 1.1263157894736843e-05,
"loss": 0.5384,
"step": 3860
},
{
"epoch": 0.64,
"learning_rate": 1.1157894736842105e-05,
"loss": 0.5415,
"step": 3880
},
{
"epoch": 0.65,
"learning_rate": 1.1052631578947368e-05,
"loss": 0.5321,
"step": 3900
},
{
"epoch": 0.65,
"learning_rate": 1.0947368421052633e-05,
"loss": 0.5342,
"step": 3920
},
{
"epoch": 0.65,
"learning_rate": 1.0842105263157895e-05,
"loss": 0.5312,
"step": 3940
},
{
"epoch": 0.66,
"learning_rate": 1.0736842105263158e-05,
"loss": 0.536,
"step": 3960
},
{
"epoch": 0.66,
"learning_rate": 1.0631578947368421e-05,
"loss": 0.5314,
"step": 3980
},
{
"epoch": 0.66,
"learning_rate": 1.0526315789473684e-05,
"loss": 0.536,
"step": 4000
},
{
"epoch": 0.67,
"learning_rate": 1.0421052631578948e-05,
"loss": 0.5336,
"step": 4020
},
{
"epoch": 0.67,
"learning_rate": 1.0315789473684211e-05,
"loss": 0.5343,
"step": 4040
},
{
"epoch": 0.67,
"learning_rate": 1.0210526315789474e-05,
"loss": 0.5287,
"step": 4060
},
{
"epoch": 0.68,
"learning_rate": 1.0105263157894738e-05,
"loss": 0.5316,
"step": 4080
},
{
"epoch": 0.68,
"learning_rate": 9.999999999999999e-06,
"loss": 0.5266,
"step": 4100
},
{
"epoch": 0.68,
"learning_rate": 9.894736842105264e-06,
"loss": 0.5249,
"step": 4120
},
{
"epoch": 0.69,
"learning_rate": 9.789473684210527e-06,
"loss": 0.5255,
"step": 4140
},
{
"epoch": 0.69,
"learning_rate": 9.68421052631579e-06,
"loss": 0.5266,
"step": 4160
},
{
"epoch": 0.69,
"learning_rate": 9.578947368421052e-06,
"loss": 0.5234,
"step": 4180
},
{
"epoch": 0.7,
"learning_rate": 9.473684210526315e-06,
"loss": 0.5264,
"step": 4200
},
{
"epoch": 0.7,
"learning_rate": 9.36842105263158e-06,
"loss": 0.532,
"step": 4220
},
{
"epoch": 0.7,
"learning_rate": 9.263157894736842e-06,
"loss": 0.5227,
"step": 4240
},
{
"epoch": 0.71,
"learning_rate": 9.157894736842105e-06,
"loss": 0.5279,
"step": 4260
},
{
"epoch": 0.71,
"learning_rate": 9.052631578947369e-06,
"loss": 0.5242,
"step": 4280
},
{
"epoch": 0.71,
"learning_rate": 8.947368421052632e-06,
"loss": 0.5195,
"step": 4300
},
{
"epoch": 0.72,
"learning_rate": 8.842105263157893e-06,
"loss": 0.5279,
"step": 4320
},
{
"epoch": 0.72,
"learning_rate": 8.736842105263158e-06,
"loss": 0.5259,
"step": 4340
},
{
"epoch": 0.72,
"learning_rate": 8.631578947368422e-06,
"loss": 0.5231,
"step": 4360
},
{
"epoch": 0.73,
"learning_rate": 8.526315789473685e-06,
"loss": 0.528,
"step": 4380
},
{
"epoch": 0.73,
"learning_rate": 8.421052631578947e-06,
"loss": 0.5194,
"step": 4400
},
{
"epoch": 0.73,
"learning_rate": 8.31578947368421e-06,
"loss": 0.5218,
"step": 4420
},
{
"epoch": 0.74,
"learning_rate": 8.210526315789475e-06,
"loss": 0.5218,
"step": 4440
},
{
"epoch": 0.74,
"learning_rate": 8.105263157894738e-06,
"loss": 0.5198,
"step": 4460
},
{
"epoch": 0.74,
"learning_rate": 8e-06,
"loss": 0.5197,
"step": 4480
},
{
"epoch": 0.75,
"learning_rate": 7.894736842105263e-06,
"loss": 0.5221,
"step": 4500
},
{
"epoch": 0.75,
"learning_rate": 7.789473684210526e-06,
"loss": 0.5198,
"step": 4520
},
{
"epoch": 0.75,
"learning_rate": 7.68421052631579e-06,
"loss": 0.5254,
"step": 4540
},
{
"epoch": 0.76,
"learning_rate": 7.578947368421053e-06,
"loss": 0.5196,
"step": 4560
},
{
"epoch": 0.76,
"learning_rate": 7.473684210526316e-06,
"loss": 0.5161,
"step": 4580
},
{
"epoch": 0.76,
"learning_rate": 7.3684210526315784e-06,
"loss": 0.5212,
"step": 4600
},
{
"epoch": 0.77,
"learning_rate": 7.2631578947368426e-06,
"loss": 0.5211,
"step": 4620
},
{
"epoch": 0.77,
"learning_rate": 7.157894736842105e-06,
"loss": 0.5162,
"step": 4640
},
{
"epoch": 0.77,
"learning_rate": 7.052631578947369e-06,
"loss": 0.52,
"step": 4660
},
{
"epoch": 0.78,
"learning_rate": 6.9473684210526315e-06,
"loss": 0.5069,
"step": 4680
},
{
"epoch": 0.78,
"learning_rate": 6.842105263157895e-06,
"loss": 0.5132,
"step": 4700
},
{
"epoch": 0.78,
"learning_rate": 6.736842105263158e-06,
"loss": 0.5174,
"step": 4720
},
{
"epoch": 0.79,
"learning_rate": 6.631578947368421e-06,
"loss": 0.5146,
"step": 4740
},
{
"epoch": 0.79,
"learning_rate": 6.526315789473685e-06,
"loss": 0.5102,
"step": 4760
},
{
"epoch": 0.79,
"learning_rate": 6.421052631578948e-06,
"loss": 0.516,
"step": 4780
},
{
"epoch": 0.8,
"learning_rate": 6.31578947368421e-06,
"loss": 0.5122,
"step": 4800
},
{
"epoch": 0.8,
"learning_rate": 6.210526315789474e-06,
"loss": 0.512,
"step": 4820
},
{
"epoch": 0.8,
"learning_rate": 6.105263157894737e-06,
"loss": 0.5128,
"step": 4840
},
{
"epoch": 0.81,
"learning_rate": 6e-06,
"loss": 0.5189,
"step": 4860
},
{
"epoch": 0.81,
"learning_rate": 5.8947368421052634e-06,
"loss": 0.5125,
"step": 4880
},
{
"epoch": 0.81,
"learning_rate": 5.789473684210526e-06,
"loss": 0.5127,
"step": 4900
},
{
"epoch": 0.82,
"learning_rate": 5.68421052631579e-06,
"loss": 0.5158,
"step": 4920
},
{
"epoch": 0.82,
"learning_rate": 5.578947368421052e-06,
"loss": 0.5124,
"step": 4940
},
{
"epoch": 0.82,
"learning_rate": 5.4736842105263165e-06,
"loss": 0.511,
"step": 4960
},
{
"epoch": 0.83,
"learning_rate": 5.368421052631579e-06,
"loss": 0.5134,
"step": 4980
},
{
"epoch": 0.83,
"learning_rate": 5.263157894736842e-06,
"loss": 0.5091,
"step": 5000
},
{
"epoch": 0.83,
"learning_rate": 5.1578947368421055e-06,
"loss": 0.5127,
"step": 5020
},
{
"epoch": 0.84,
"learning_rate": 5.052631578947369e-06,
"loss": 0.5075,
"step": 5040
},
{
"epoch": 0.84,
"learning_rate": 4.947368421052632e-06,
"loss": 0.5131,
"step": 5060
},
{
"epoch": 0.84,
"learning_rate": 4.842105263157895e-06,
"loss": 0.5103,
"step": 5080
},
{
"epoch": 0.85,
"learning_rate": 4.736842105263158e-06,
"loss": 0.5087,
"step": 5100
},
{
"epoch": 0.85,
"learning_rate": 4.631578947368421e-06,
"loss": 0.5105,
"step": 5120
},
{
"epoch": 0.85,
"learning_rate": 4.526315789473684e-06,
"loss": 0.5069,
"step": 5140
},
{
"epoch": 0.86,
"learning_rate": 4.421052631578947e-06,
"loss": 0.5152,
"step": 5160
},
{
"epoch": 0.86,
"learning_rate": 4.315789473684211e-06,
"loss": 0.5118,
"step": 5180
},
{
"epoch": 0.86,
"learning_rate": 4.210526315789473e-06,
"loss": 0.5138,
"step": 5200
},
{
"epoch": 0.87,
"learning_rate": 4.105263157894737e-06,
"loss": 0.5112,
"step": 5220
},
{
"epoch": 0.87,
"learning_rate": 4e-06,
"loss": 0.5069,
"step": 5240
},
{
"epoch": 0.87,
"learning_rate": 3.894736842105263e-06,
"loss": 0.5086,
"step": 5260
},
{
"epoch": 0.88,
"learning_rate": 3.7894736842105264e-06,
"loss": 0.5102,
"step": 5280
},
{
"epoch": 0.88,
"learning_rate": 3.6842105263157892e-06,
"loss": 0.5105,
"step": 5300
},
{
"epoch": 0.88,
"learning_rate": 3.5789473684210525e-06,
"loss": 0.5072,
"step": 5320
},
{
"epoch": 0.89,
"learning_rate": 3.4736842105263158e-06,
"loss": 0.5076,
"step": 5340
},
{
"epoch": 0.89,
"learning_rate": 3.368421052631579e-06,
"loss": 0.509,
"step": 5360
},
{
"epoch": 0.89,
"learning_rate": 3.2631578947368423e-06,
"loss": 0.5069,
"step": 5380
},
{
"epoch": 0.9,
"learning_rate": 3.157894736842105e-06,
"loss": 0.5042,
"step": 5400
},
{
"epoch": 0.9,
"learning_rate": 3.0526315789473684e-06,
"loss": 0.5134,
"step": 5420
},
{
"epoch": 0.9,
"learning_rate": 2.9473684210526317e-06,
"loss": 0.502,
"step": 5440
},
{
"epoch": 0.91,
"learning_rate": 2.842105263157895e-06,
"loss": 0.5044,
"step": 5460
},
{
"epoch": 0.91,
"learning_rate": 2.7368421052631583e-06,
"loss": 0.5102,
"step": 5480
},
{
"epoch": 0.91,
"learning_rate": 2.631578947368421e-06,
"loss": 0.5061,
"step": 5500
},
{
"epoch": 0.92,
"learning_rate": 2.5263157894736844e-06,
"loss": 0.4996,
"step": 5520
},
{
"epoch": 0.92,
"learning_rate": 2.4210526315789477e-06,
"loss": 0.5049,
"step": 5540
},
{
"epoch": 0.92,
"learning_rate": 2.3157894736842105e-06,
"loss": 0.5061,
"step": 5560
},
{
"epoch": 0.93,
"learning_rate": 2.2105263157894734e-06,
"loss": 0.5041,
"step": 5580
},
{
"epoch": 0.93,
"learning_rate": 2.1052631578947366e-06,
"loss": 0.5021,
"step": 5600
},
{
"epoch": 0.93,
"learning_rate": 2e-06,
"loss": 0.5033,
"step": 5620
},
{
"epoch": 0.94,
"learning_rate": 1.8947368421052632e-06,
"loss": 0.5013,
"step": 5640
},
{
"epoch": 0.94,
"learning_rate": 1.7894736842105262e-06,
"loss": 0.5053,
"step": 5660
},
{
"epoch": 0.94,
"learning_rate": 1.6842105263157895e-06,
"loss": 0.5026,
"step": 5680
},
{
"epoch": 0.95,
"learning_rate": 1.5789473684210526e-06,
"loss": 0.5042,
"step": 5700
},
{
"epoch": 0.95,
"learning_rate": 1.4736842105263159e-06,
"loss": 0.5026,
"step": 5720
},
{
"epoch": 0.95,
"learning_rate": 1.3684210526315791e-06,
"loss": 0.5021,
"step": 5740
},
{
"epoch": 0.96,
"learning_rate": 1.2631578947368422e-06,
"loss": 0.4996,
"step": 5760
},
{
"epoch": 0.96,
"learning_rate": 1.1578947368421053e-06,
"loss": 0.503,
"step": 5780
},
{
"epoch": 0.96,
"learning_rate": 1.0526315789473683e-06,
"loss": 0.4996,
"step": 5800
},
{
"epoch": 0.97,
"learning_rate": 9.473684210526316e-07,
"loss": 0.5038,
"step": 5820
},
{
"epoch": 0.97,
"learning_rate": 8.421052631578948e-07,
"loss": 0.5025,
"step": 5840
},
{
"epoch": 0.97,
"learning_rate": 7.368421052631579e-07,
"loss": 0.5025,
"step": 5860
},
{
"epoch": 0.98,
"learning_rate": 6.315789473684211e-07,
"loss": 0.5061,
"step": 5880
},
{
"epoch": 0.98,
"learning_rate": 5.263157894736842e-07,
"loss": 0.4971,
"step": 5900
},
{
"epoch": 0.98,
"learning_rate": 4.210526315789474e-07,
"loss": 0.505,
"step": 5920
},
{
"epoch": 0.99,
"learning_rate": 3.1578947368421055e-07,
"loss": 0.4992,
"step": 5940
},
{
"epoch": 0.99,
"learning_rate": 2.105263157894737e-07,
"loss": 0.4982,
"step": 5960
},
{
"epoch": 0.99,
"learning_rate": 1.0526315789473685e-07,
"loss": 0.5003,
"step": 5980
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.5002,
"step": 6000
}
],
"max_steps": 6000,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}