| { | |
| "_name_or_path": "codesage/codesage-small", | |
| "activation_function": "gelu_new", | |
| "architectures": [ | |
| "CodeSageModel" | |
| ], | |
| "attention_dropout_prob": 0.1, | |
| "auto_map": { | |
| "AutoConfig": "codesage/codesage-small--config_codesage.CodeSageConfig", | |
| "AutoModel": "codesage/codesage-small--modeling_codesage.CodeSageModel", | |
| "AutoModelForMaskedLM": "codesage/codesage-small--modeling_codesage.CodeSageForMaskedLM", | |
| "AutoModelForSequenceClassification": "codesage/codesage-small--modeling_codesage.CodeSageForSequenceClassification" | |
| }, | |
| "bos_token_id": 0, | |
| "embedding_dropout_prob": 0.1, | |
| "eos_token_id": 0, | |
| "hidden_size": 1024, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_norm_epsilon": 1e-05, | |
| "max_position_embeddings": 2048, | |
| "model_type": "codesage", | |
| "num_attention_heads": 8, | |
| "num_hidden_layers": 6, | |
| "pad_token_id": 49153, | |
| "position_embedding_type": "absolute", | |
| "residual_dropout_prob": 0.1, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.36.2", | |
| "vocab_size": 49154 | |
| } | |