| { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "model_type": "llama", | |
| "hidden_size": 1024, | |
| "num_hidden_layers": 48, | |
| "num_attention_heads": 16, | |
| "num_key_value_heads": 4, | |
| "intermediate_size": 2816, | |
| "vocab_size": 65536, | |
| "max_position_embeddings": 2048, | |
| "rope_theta": 10000, | |
| "tie_word_embeddings": true, | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "hidden_act": "silu", | |
| "initializer_range": 0.02, | |
| "rms_norm_eps": 1e-06, | |
| "mlp_bias": false, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true | |
| } | |