Preprint
Aug 2026
ExFold: Unified Expert Folding for Training-Free MoE Prefill-Decode Acceleration
ExFold is proposed, a unified training-free expert-folding framework for jointly accelerating MoE prefill and decode that calibrates a pairwise scalar-projector matrix on unlabeled data and uses it at inference time to fold excluded expert contributions into retained experts.
Jun-Tong Wu, Yi-Fei Liu, Jun-Yi Chen et al.
· 2 citations
· ⚡2