{
  "architectures": ["LlamaForCausalLM"],
  "model_type": "llama",
  "hidden_size": 4096,
  "intermediate_size": 11008,
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "max_position_embeddings": 2048,
  "vocab_size": 32000,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "use_cache": true
}