config.json

{
    "architectures": ["LLaMAModel"],  
    "model_type": "llama",
    "hidden_size": 4096,  
    "num_hidden_layers": 30,  
    "num_attention_heads": 16,  
    "intermediate_size": 11008,  
    "max_position_embeddings": 8192, 
    "vocab_size": 128256, 
    "activation_function": "gelu_new", 
    "layer_norm_eps": 1e-5,
    "initializer_range": 0.02,
    "bos_token_id": 128000,  
    "eos_token_id": 128001, 
    "pad_token_id": 128255,  
    "special_tokens_map": {
        "bos_token": "<|begin_of_text|>",
        "eos_token": "<|end_of_text|>",
        "pad_token": "<|reserved_special_token_250|>"
    },
    "model_input_names": ["input_ids", "attention_mask"],
    "task_specific_params": {
        "text-generation": {
            "do_sample": true,
            "max_length": 50
        }
    }
}