dSpark：面向快速扩散语言模型的调度蒸馏方法，学习何时提交 token
Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models
AI 导读
提出基于调度蒸馏的 dSpark 框架，通过构建优先揭示能降低其他位置不确定性的 token 的解掩码调度，并以 KL 散度正则项蒸馏进模型，实现更快的并行采样。在 GSM8K 上，该方法在 LLaDA-8B-Instruct 和 Dream-7B-Instruct 上分别取得 10.12× 和 5.12× 加速，同时保持更优的任务表现。代码已开源。
正文 · 原文
Abstract:Diffusion large language models (dLLMs) can generate multiple tokens in parallel, but often require many decoding rounds to complete a response. Efficient generation depends not only on which tokens are easy to predict, but also on how committing to them helps predict the remaining positions. A highly predictable token may provide little useful context, while revealing a less predictable token can help several other positions become predictable sooner. We propose dSpark, a new schedule distillation-based framework for learning when to commit tokens for faster diffusion language generation. To achieve this, we construct unmasking schedules that prioritize tokens whose reveal makes other positions easier to predict, reducing uncertainty earlier in decoding. We then distill these schedules into the model to enable faster parallel sampling, with a KL-divergence regularizer to help preserve generation quality. We extensively evaluate our approach across mathematical reasoning and code generation tasks. On GSM8K, our method achieves remarkable 10.12$\times$ and 5.12$\times$ speedups over the various baselines on LLaDA-8B-Instruct and Dream-7B-Instruct, respectively, while maintaining superior task performance. Our code is available at this https URL.
| Comments: | Code is available at this https URL |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.35804 [cs.CL] |
| (or arXiv:2610.35804v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35804 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Aidar Myrzakhan [view email]
[v1]
Mon, 26 Oct 2026 17:35:58 UTC (523 KB)
来源：arXiv cs.AI · arxiv.org