| { | |
| "architectures": [ | |
| "CompactPhi3Model" | |
| ], | |
| "attention_dropout": 0.0, | |
| "attention_input_size": 1536, | |
| "attention_output_size": 3072, | |
| "embd_pdrop": 0.0, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 4096, | |
| "mlp_hidden_input_size": 1536, | |
| "mlp_hidden_output_size": 3072, | |
| "mlp_intermediate_size": 4096, | |
| "model_type": "CompactPhi3Model", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 32, | |
| "original_hidden_size": 3072, | |
| "original_max_position_embeddings": 4096, | |
| "resid_pdrop": 0.0, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "sliding_window": null, | |
| "substituted_layers": [ | |
| 15, | |
| 19, | |
| 23, | |
| 27 | |
| ], | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.41.1", | |
| "use_cache": true, | |
| "vocab_size": 32064 | |
| } | |