| { | |
| "architectures": ["Qwen2ForCausalLM"], | |
| "model_type": "qwen2", | |
| "hidden_size": 1024, | |
| "intermediate_size": 3072, | |
| "num_hidden_layers": 28, | |
| "num_attention_heads": 16, | |
| "num_key_value_heads": 8, | |
| "head_dim": 128, | |
| "vocab_size": 151936, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 1000000.0, | |
| "max_position_embeddings": 40960, | |
| "hidden_act": "silu", | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true, | |
| "tie_word_embeddings": false | |
| } | |