{ "activation": "relu", "architectures": [ "DistilBertForMaskedLM" ], "attention_dropout": 0.05, "dim": 64, "dropout": 0.05, "hidden_dim": 256, "initializer_range": 0.02, "max_position_embeddings": 512, "model_type": "hybridbert", "n_heads": 2, "n_layers": 10, "pad_token_id": 0, "qa_dropout": 0.1, "seq_classif_dropout": 0.1, "sinusoidal_pos_embds": true, "torch_dtype": "float32", "transformers_version": "4.33.3", "vocab_size": 30522 }