{ "dataset_reader": { "type": "ccqa", "lazy": true, "token_indexers": { "bert": { "type": "pretrained_transformer", "model_name": "roberta-base" } }, "tokenizer": { "type": "pretrained_transformer", "add_special_tokens": false, "model_name": "roberta-base" } }, "model": { "type": "basic_classifier_modified", "seq2vec_encoder": { "type": "bert_pooler", "override_weights_file": "PretrainedRoberta/checkpoint_best.pt", "pretrained_model": "roberta-base", "requires_grad": true }, "text_field_embedder": { "type": "basic", "token_embedders": { "bert": { "type": "pretrained_transformer", "last_layer_only": false, "model_name": "roberta-base", "override_weights_file": "PretrainedRoberta/checkpoint_best.pt" } } } }, "train_data_path": "SufficientCompanyData/train_df_balanced_1500_15000.tsv", "validation_data_path": "SufficientCompanyData/val_df_balanced_750_7500.tsv", "trainer": { "checkpointer": { "num_serialized_models_to_keep": 1 }, "cuda_device": 0, "grad_norm": 1, "num_epochs": 50, "optimizer": { "type": "huggingface_adamw", "lr": 2e-06, "weight_decay": 0.1 }, "validation_metric": "+accuracy" }, "data_loader": { "batch_size": 8 } }