config.json

{
  "architectures": [
    "MistralForCausalLM"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 5120,
  "initializer_range": 0.02,
  "intermediate_size": 32768,
  "max_position_embeddings": 32768,
  "model_type": "mistral",
  "num_attention_heads": 32,
  "num_hidden_layers": 40,
  "num_key_value_heads": 8,
  "rms_norm_eps": 1e-05,
  "rope_theta": 1000000000.0,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.52.4",
  "use_cache": true,
  "vocab_size": 131072,
  "quantization_config": {
    "quant_method": "fp8",
    "is_checkpoint_fp8_serialized": true,
    "activation_scheme": "dynamic",
    "weight_block_size": null,
    "ignored_layers": [
      "lm_head"
    ]
  }
}