| { | |
| "_name_or_path": "../fusing-models/bert/", | |
| "activation_dropout": 0.0, | |
| "activation_function": "gelu", | |
| "architectures": [ | |
| "LDMBertModel" | |
| ], | |
| "attention_dropout": 0.0, | |
| "classifier_dropout": 0.0, | |
| "d_model": 1280, | |
| "dropout": 0.1, | |
| "encoder_attention_heads": 8, | |
| "encoder_ffn_dim": 5120, | |
| "encoder_layerdrop": 0.0, | |
| "encoder_layers": 32, | |
| "head_dim": 64, | |
| "init_std": 0.02, | |
| "max_position_embeddings": 77, | |
| "model_type": "ldmbert", | |
| "num_hidden_layers": 32, | |
| "pad_token_id": 0, | |
| "scale_embedding": false, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.20.0.dev0", | |
| "use_cache": true, | |
| "vocab_size": 30522 | |
| } | |