| { | |
| "architectures": [ | |
| "LLM" | |
| ], | |
| "model_type": "llm", | |
| "vocab_size": 131072, | |
| "hidden_size": 2048, | |
| "num_hidden_layers": 16, | |
| "num_attention_heads": 32, | |
| "num_key_value_heads": 8, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 4096, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 500000, | |
| "tie_word_embeddings": true, | |
| "attention_bias": false, | |
| "mlp_bias": false | |
| } |