config.json

{
  "architectures": ["MistralForCausalLM"],
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_size": 4096,
  "intermediate_size": 11008,
  "max_position_embeddings": 4096,
  "model_type": "mistral",
  "n_embd": 4096,
  "n_head": 32,
  "n_layer": 32,
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "pad_token_id": 0,
  "tokenizer_class": "MistralTokenizer",
  "vocab_size": 32000,
  "gradient_checkpointing": true,
  "use_cache": false
}