| { |
| "_name_or_path": "/weight/llm/microsoft/phi-4", |
| "architectures": [ |
| "Phi3ForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "auto_map": {}, |
| "bos_token_id": 100257, |
| "embd_pdrop": 0.0, |
| "eos_token_id": 100265, |
| "hidden_act": "silu", |
| "hidden_size": 5120, |
| "initializer_range": 0.02, |
| "intermediate_size": 17920, |
| "max_position_embeddings": 16384, |
| "model_type": "phi3", |
| "num_attention_heads": 40, |
| "num_hidden_layers": 40, |
| "num_key_value_heads": 10, |
| "original_max_position_embeddings": 16384, |
| "pad_token_id": 100349, |
| "quantization_config": { |
| "bits": 4, |
| "group_size": 128, |
| "modules_to_not_convert": null, |
| "quant_method": "awq", |
| "version": "gemm", |
| "zero_point": true |
| }, |
| "resid_pdrop": 0.0, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": null, |
| "rope_theta": 250000, |
| "sliding_window": null, |
| "tie_word_embeddings": false, |
| "torch_dtype": "bfloat16", |
| "transformers_version": "4.47.1", |
| "use_cache": false, |
| "vocab_size": 100352 |
| } |
|
|