{
  "model_type": "llama",
  "architectures": [
    "LlamaForCausalLM"
  ],
  "num_attention_heads": 32,
  "hidden_size": 4096,
  "intermediate_size": 11008,
  "max_position_embeddings": 2048,
  "from_tf": true,
  "torch_dtype": "float16",
  "pad_token_id": 0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "tie_word_embeddings": false
}