Firworks
/

MiniMax-M2.1-nvfp4

compressed-tensors

Model card Files Files and versions

MiniMax-M2.1-nvfp4 / recipe.yaml

Firworks's picture

Upload folder using huggingface_hub

83c07c3 verified 3 days ago

history blame contribute delete

199 Bytes

	default_stage:
	default_modifiers:
	QuantizationModifier:
	targets: [Linear]
	ignore: [lm_head, 're:visual.', 're:.vision_tower.', 're:.gate$', 're:.router.']
	scheme: NVFP4