hf_quant_config.json

{
    "producer": {
        "name": "modelopt",
        "version": "0.37.0.dev5+g76fb12d47.d20250905"
    },
    "quantization": {
        "quant_algo": "FP8",
        "kv_cache_quant_algo": null,
        "exclude_modules": [
            "model.layers.language_model.lm_head",
            "model.layers.mlp1*",
	        "model.layers.*.conv1d*",
            "model.layers.vision_model*",
            "lm_head"
        ]
    }
}