thwin27
/

Aria-sequential_mlp-FP8-dynamic

Image-Text-to-Text

text-generation

compressed-tensors

Model card Files Files and versions Community

Aria-sequential_mlp-FP8-dynamic / recipe.yaml

thwin27's picture

Upload folder using huggingface_hub

e79c572 verified 11 days ago

192 Bytes

	DEFAULT_stage:
	DEFAULT_modifiers:
	QuantizationModifier:
	ignore: ['re:.lm_head', 're:multi_modal_projector.', 're:vision_tower.*']
	targets: Linear
	scheme: FP8_DYNAMIC