| { | |
| "batching": { | |
| "batch_size": 64, | |
| "max_tokens": 1024, | |
| "sampling_smoothing": 1, | |
| "sort_by_size": true | |
| }, | |
| "decoders": { | |
| "classification": {}, | |
| "default_decoder": { | |
| "layers_to_use": [ | |
| -1 | |
| ], | |
| "loss_weight": 1, | |
| "metric": "accuracy", | |
| "topn": 1 | |
| }, | |
| "dependency": { | |
| "arc_representation_dim": 768, | |
| "metric": "las", | |
| "tag_representation_dim": 256 | |
| }, | |
| "mlm": { | |
| "metric": "perplexity" | |
| }, | |
| "multiclas": { | |
| "metric": "multi_acc", | |
| "threshold": 0.7 | |
| }, | |
| "multiseq": { | |
| "metric": "multi_acc", | |
| "threshold": 0.7 | |
| }, | |
| "regression": { | |
| "metric": "avg_dist" | |
| }, | |
| "seq": {}, | |
| "seq_bio": { | |
| "metric": "span_f1" | |
| }, | |
| "string2string": {}, | |
| "tok": { | |
| "pre_split": true | |
| } | |
| }, | |
| "default_dec_dataset_embeds_dim": 12, | |
| "encoder": { | |
| "dropout": 0.2, | |
| "max_input_length": 64, | |
| "update_weights_encoder": true | |
| }, | |
| "random_seed": 8446, | |
| "training": { | |
| "keep_top_n": 1, | |
| "learning_rate_scheduler": { | |
| "cut_frac": 0.3, | |
| "decay_factor": 0.38, | |
| "discriminative_fine_tuning": true, | |
| "gradual_unfreezing": true | |
| }, | |
| "num_epochs": 50, | |
| "optimizer": { | |
| "betas": [ | |
| 0.9, | |
| 0.99 | |
| ], | |
| "correct_bias": false, | |
| "lr": 0.0015, | |
| "weight_decay": 0.01 | |
| } | |
| }, | |
| "transformer_model": "dbmdz/bert-base-italian-xxl-cased" | |
| } |