YFarmX

Weight-only quantisation

AI

Weight-only quantisation: Compressing just the model's stored weights to low precision while keeping activations at higher precision, saving memory with minimal effect on quality.

Related terms

Browse the full glossary →