| { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "model_type": "llama", | |
| "hidden_size": 2048, | |
| "intermediate_size": 5472, | |
| "num_hidden_layers": 24, | |
| "num_attention_heads": 16, | |
| "num_key_value_heads": 4, | |
| "hidden_act": "silu", | |
| "max_position_embeddings": 4096, | |
| "initializer_range": 0.02, | |
| "rms_norm_eps": 1e-05, | |
| "vocab_size": 64256, | |
| "rope_theta": 500000.0, | |
| "rope_scaling": null, | |
| "attention_bias": false, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.40.0" | |
| } |