{ "_name_or_path": "sentence-transformers/msmarco-distilbert-base-v2", "activation": "gelu", "architectures": [ "BertForSequenceClassification" ], "attention_dropout": 0.1, "attention_probs_dropout_prob": 0.1, "classifier_dropout": null, "dim": 768, "dropout": 0.1, "hidden_act": "gelu", "hidden_dim": 3072, "hidden_dropout_prob": 0.1, "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "layer_norm_eps": 1e-12, "max_position_embeddings": 512, "model_type": "bert", "n_heads": 12, "n_layers": 6, "num_attention_heads": 12, "num_hidden_layers": 12, "pad_token_id": 0, "position_embedding_type": "absolute", "problem_type": "single_label_classification", "qa_dropout": 0.1, "seq_classif_dropout": 0.2, "sinusoidal_pos_embds": false, "tie_weights_": true, "torch_dtype": "float32", "transformers_version": "4.16.2", "type_vocab_size": 2, "use_cache": true, "vocab_size": 30522 }