{
    "quantization": {
        "quant_algo": "FP8",
        "kv_cache_quant_algo": null,
        "exclude_modules": [
            "lm_head",
            "model.embed_tokens"
        ]
    }
}