| { | |
| "model_type": "qwen", | |
| "architectures": ["QwenForCausalLM"], | |
| "use_safetensors": true, | |
| "vocab_size": 151936, | |
| "hidden_size": 4096, | |
| "num_hidden_layers": 32, | |
| "num_attention_heads": 32, | |
| "intermediate_size": 11008, | |
| "hidden_act": "gelu", | |
| "max_position_embeddings": 2048, | |
| "initializer_range": 0.02, | |
| "layer_norm_eps": 1e-5, | |
| "use_cache": true, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.36.0", | |
| "pad_token_id": 151643, | |
| "bos_token_id": 151643, | |
| "eos_token_id": 151643, | |
| "attention_dropout": 0.1, | |
| "hidden_dropout": 0.1, | |
| "num_key_value_heads": 32, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": false | |
| } |