{
  "model_type": "llama",
  "architectures": [
    "LLaMAForCausalLM"
  ],
  "hidden_size": 2048,
  "intermediate_size": 5632,
  "max_position_embeddings": 2048,
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "vocab_size": 32000,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "pad_token_id": 0,
  "torch_dtype": "float16"
}