Saved in:
Bibliographic Details
Main Authors: Wang, Zhuoyu, Huang, Junnan, Chen, Xinyu
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2606.00487
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913175051436032
author Wang, Zhuoyu
Huang, Junnan
Chen, Xinyu
author_facet Wang, Zhuoyu
Huang, Junnan
Chen, Xinyu
contents Using a diffusion model for parallel drafting is a promising approach for speculative decoding. By predicting tokens at multiple future positions in a single forward pass, diffusion drafters substantially reduce drafting latency. However, this shifts the bottleneck to verification: verifying a single sequence limits acceptance length, while verifying large draft trees incurs excessive target-model latency. We identify a key mismatch in existing draft-tree methods: existing diffusion-tree methods rank nodes by the marginal probability, ignoring that verification is prefix-conditioned. As a result, they may verify unreachable descendants of rejected prefixes, increasing latency with limited acceptance gains. To address this, we propose TAPS, a target-aware prefix selection method that turns diffusion marginals into path-conditioned acceptance estimates. TAPS then selects a compact prefix-closed subtree under a fixed verification budget, improving the acceptance-cost tradeoff rather than simply expanding the draft tree. Experiments across diverse datasets and model families demonstrate that TAPS achieves up to 7.9x lossless end-to-end speedup over vanilla autoregressive decoding, outperforming state-of-the-art DFlash and DDTree by 1.36x and 1.74x respectively. Our work is available at https://anonymous.4open.science/r/TAPS-EMNLP2026-53DD
format Preprint
id arxiv_https___arxiv_org_abs_2606_00487
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TAPS: Target-Aware Prefix Tree Selection for Diffusion-Drafted Speculative Decoding
Wang, Zhuoyu
Huang, Junnan
Chen, Xinyu
Artificial Intelligence
Using a diffusion model for parallel drafting is a promising approach for speculative decoding. By predicting tokens at multiple future positions in a single forward pass, diffusion drafters substantially reduce drafting latency. However, this shifts the bottleneck to verification: verifying a single sequence limits acceptance length, while verifying large draft trees incurs excessive target-model latency. We identify a key mismatch in existing draft-tree methods: existing diffusion-tree methods rank nodes by the marginal probability, ignoring that verification is prefix-conditioned. As a result, they may verify unreachable descendants of rejected prefixes, increasing latency with limited acceptance gains. To address this, we propose TAPS, a target-aware prefix selection method that turns diffusion marginals into path-conditioned acceptance estimates. TAPS then selects a compact prefix-closed subtree under a fixed verification budget, improving the acceptance-cost tradeoff rather than simply expanding the draft tree. Experiments across diverse datasets and model families demonstrate that TAPS achieves up to 7.9x lossless end-to-end speedup over vanilla autoregressive decoding, outperforming state-of-the-art DFlash and DDTree by 1.36x and 1.74x respectively. Our work is available at https://anonymous.4open.science/r/TAPS-EMNLP2026-53DD
title TAPS: Target-Aware Prefix Tree Selection for Diffusion-Drafted Speculative Decoding
topic Artificial Intelligence
url https://arxiv.org/abs/2606.00487