Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding

published

Abstract

No abstract available.

Cite this paper

@article{lehan2026,
  title = {Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding},
  author = {Pan, Lehan and Tao, Ziyang and Wang, Xiao and Zhao, Zhimeng and Wang, Meng and Zhang, Yanyong},
  journal = {Findings of the Association for Computational Linguistics: EMNLP 2026},
  year = {2026},
  month = {Jan},
}