anhbn
/

Phi-3.5-mini-instruct-quantized.w8a8

8-bit precision

compressed-tensors

Model card Files Files and versions Community

Phi-3.5-mini-instruct-quantized.w8a8 / recipe.yaml

anhbn's picture

Upload folder using huggingface_hub

b281c7c verified 19 days ago

history blame contribute delete

146 Bytes

	DEFAULT_stage:
	DEFAULT_modifiers:
	GPTQModifier:
	targets: Linear
	dampening_frac: 0.01
	ignore: [lm_head]
	scheme: W8A8