RedHatAI
/

Voxtral-Mini-3B-2507-FP8-dynamic

Automatic Speech Recognition

text-generation-inference

automatic-speech-translation

audio-text-to-text

video-text-to-text

compressed-tensors

Model card Files Files and versions

Voxtral-Mini-3B-2507-FP8-dynamic / recipe.yaml

alexmarques's picture

Upload folder using huggingface_hub

ec72c57 verified 8 months ago

history blame contribute delete

201 Bytes

	default_stage:
	default_modifiers:
	QuantizationModifier:
	targets: [Linear]
	ignore: [language_model.lm_head, 're:audio_tower.', 're:multi_modal_projector.']
	scheme: FP8_DYNAMIC