| { | |
| "architectures": ["Qwen3ForCausalLM"], | |
| "model_type": "qwen3", | |
| "hidden_size": 2560, | |
| "intermediate_size": 9728, | |
| "num_hidden_layers": 36, | |
| "num_attention_heads": 32, | |
| "num_key_value_heads": 8, | |
| "head_dim": 128, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 5000000, | |
| "rope_scaling": null, | |
| "sliding_window": null, | |
| "max_position_embeddings": 262144, | |
| "max_window_layers": 36, | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 16, | |
| "padded_vocab_size": 16, | |
| "bos_token_id": 14, | |
| "eos_token_id": 15 | |
| } |