nm-testing
/

granite-20b-code-instruct-8k-quantized.w4a16

compressed-tensors

Model card Files Files and versions

granite-20b-code-instruct-8k-quantized.w4a16 / recipe.yaml

kylesayrs's picture

Upload folder using huggingface_hub

9072571 verified 10 months ago

history blame contribute delete

275 Bytes

	default_stage:
	default_modifiers:
	GPTQModifier:
	targets: [Linear]
	ignore: [lm_head]
	scheme: W4A16
	sequential_update: true
	sequential_targets: [GPTBigCodeBlock]
	block_size: 128
	dampening_frac: 0.01
	offload_hessians: false