hf_quant_config.json

{
    "producer": {
        "name": "modelopt",
        "version": "0.42.0rc1.dev9+ge53ca61b7"
    },
    "quantization": {
        "quant_algo": "FP8",
        "kv_cache_quant_algo": "FP8",
        "exclude_modules": [
            "lm_head",
            "model.layers.92*"
        ]
    }
}