Revisiting Lossy Verification in Speculative Decoding: Mechanisms, Trade-offs, and Failure Modes
Tianyu Wang, Yuxuan Zhou, Wenbin Wang, Heng Li, Zikai Xiao, Junyuan Shang
This paper studies how lossy verification schemes can improve the efficiency of speculative decoding in large language models, but may also degrade generation quality. Practitioners may care about understanding the trade-offs between speed and quality when using these schemes.
Abstract
Speculative Decoding (SD) accelerates large language model inference by allowing a lightweight draft model to propose tokens that are subsequently verified in parallel by a larger target model. Recent approaches introduce lossy verification schemes to further improve efficiency by relaxing strict distributional matching. Yet such relaxation silently rewrites the decoding distribution, and the resulting acceleration can come at the cost of unstable, sometimes severely degraded generation quality. In this work, we present a principled analysis of the distributions induced by lossy verification methods. We show that many seemingly distinct approaches differ only superficially and can be classified into two categories: truncation-based verification and collaborative verification. We further construct a diagnostic evaluation framework across curated benchmarks. For truncation-based methods, we identify a fundamental pitfall: performance can degrade significantly compared to the true truncation sampling baseline due to distributional distortion. For collaborative verification, we uncover a key principles: controlling the overshoot of draft probabilities relative to target probabilities is essential to prevent low-quality outputs. Our code is available at https://github.com/ZhouYuxuanYX/Fast-HSD.