webbigdata
/

VoiceCore_smoothquant

8-bit precision

compressed-tensors

Model card Files Files and versions

VoiceCore_smoothquant / recipe.yaml

dahara1's picture

Upload 6 files

8c660d1 verified 4 months ago

history blame contribute delete

200 Bytes

	default_stage:
	default_modifiers:
	SmoothQuantModifier: {smoothing_strength: 0.5}
	QuantizationModifier:
	targets: [Linear]
	ignore: [lm_head, model.embed_tokens]
	scheme: W8A8