recipe.yaml

default_stage:
  default_modifiers:
    QuantizationModifier:
      targets: [Linear]
      ignore: ['re:.*lm_head', 're:.*router', 're:visual.*', 're:model.visual.*', 're:.*mlp.gate$',
        're:.*mlp.shared_expert_gate$', 're:multi_modal_projector.*']
      scheme: FP8_BLOCK