{
  "architectures": [
    "LlamaForCausalLM"
  ],
  "model_type": "llama",
  "vocab_size": 50000,
  "hidden_size": 2048,
  "num_hidden_layers": 4,
  "num_attention_heads": 16,
  "intermediate_size": 8192,
  "max_position_embeddings": 2048,
  "rms_norm_eps": 1e-05,
  "rope_theta": 10000.0,
  "tie_word_embeddings": true,
  "use_cache": true,
  "torch_dtype": "float16"
}