| { | |
| "vocab_size": 32000, | |
| "hidden_size": 384, | |
| "intermediate_size": 1536, | |
| "num_attention_heads": 8, | |
| "num_key_value_heads": 8, | |
| "num_hidden_layers": 6, | |
| "max_position_embeddings": 256, | |
| "rms_norm_eps": 1E-05, | |
| "rope_theta": 10000, | |
| "model_type": "llama", | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "torch_dtype": "float32", | |
| "hidden_act": "silu" | |
| } |