| { | |
| "dataset_reader": { | |
| "type": "coref", | |
| "max_sentences": 110, | |
| "max_span_width": 30, | |
| "token_indexers": { | |
| "tokens": { | |
| "type": "pretrained_transformer_mismatched", | |
| "max_length": 512, | |
| "model_name": "SpanBERT/spanbert-large-cased" | |
| } | |
| } | |
| }, | |
| "model": { | |
| "type": "coref", | |
| "antecedent_feedforward": { | |
| "activations": "relu", | |
| "dropout": 0.3, | |
| "hidden_dims": 1500, | |
| "input_dim": 9296, | |
| "num_layers": 2 | |
| }, | |
| "coarse_to_fine": true, | |
| "context_layer": { | |
| "type": "pass_through", | |
| "input_dim": 1024 | |
| }, | |
| "feature_size": 20, | |
| "inference_order": 2, | |
| "initializer": { | |
| "regexes": [ | |
| [ | |
| ".*_span_updating_gated_sum.*weight", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| ".*linear_layers.*weight", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| ".*scorer.*weight", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| "_distance_embedding.weight", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| "_span_width_embedding.weight", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| "_context_layer._module.weight_ih.*", | |
| { | |
| "type": "xavier_normal" | |
| } | |
| ], | |
| [ | |
| "_context_layer._module.weight_hh.*", | |
| { | |
| "type": "orthogonal" | |
| } | |
| ] | |
| ] | |
| }, | |
| "max_antecedents": 50, | |
| "max_span_width": 30, | |
| "mention_feedforward": { | |
| "activations": "relu", | |
| "dropout": 0.3, | |
| "hidden_dims": 1500, | |
| "input_dim": 3092, | |
| "num_layers": 2 | |
| }, | |
| "spans_per_word": 0.4, | |
| "text_field_embedder": { | |
| "token_embedders": { | |
| "tokens": { | |
| "type": "pretrained_transformer_mismatched", | |
| "max_length": 512, | |
| "model_name": "SpanBERT/spanbert-large-cased" | |
| } | |
| } | |
| } | |
| }, | |
| "train_data_path": "/home/dirkg/tank/data/conll12/train.english.v4_gold_conll", | |
| "validation_data_path": "/home/dirkg/tank/data/conll12/dev.english.v4_gold_conll", | |
| "test_data_path": "/home/dirkg/tank/data/conll12/test.english.v4_gold_conll", | |
| "trainer": { | |
| "learning_rate_scheduler": { | |
| "type": "slanted_triangular", | |
| "cut_frac": 0.06 | |
| }, | |
| "num_epochs": 40, | |
| "optimizer": { | |
| "type": "huggingface_adamw", | |
| "lr": 0.0003, | |
| "parameter_groups": [ | |
| [ | |
| [ | |
| ".*transformer.*" | |
| ], | |
| { | |
| "lr": 1e-05 | |
| } | |
| ] | |
| ] | |
| }, | |
| "patience": 10, | |
| "validation_metric": "+coref_f1" | |
| }, | |
| "data_loader": { | |
| "batch_sampler": { | |
| "type": "bucket", | |
| "batch_size": 1, | |
| "sorting_keys": [ | |
| "text" | |
| ] | |
| } | |
| }, | |
| "validation_dataset_reader": { | |
| "type": "coref", | |
| "max_span_width": 30, | |
| "token_indexers": { | |
| "tokens": { | |
| "type": "pretrained_transformer_mismatched", | |
| "max_length": 512, | |
| "model_name": "SpanBERT/spanbert-large-cased" | |
| } | |
| } | |
| } | |
| } |