{
  "architectures": [
    "LlamaForCausalLM"
  ],
  "model_type": "llama",
  "torch_dtype": "bfloat16",
  "transformers_version": "4.40.1",
  "hidden_size": 5120,
  "intermediate_size": 13824,
  "num_attention_heads": 40,
  "num_hidden_layers": 40,
  "num_key_value_heads": 8,
  "vocab_size": 151665
}