arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.00286cs.LG
收藏

生成顺序与并行解码在掩码扩散模型中的研究:信息论视角

Generation Order and Parallel Decoding in Masked Diffusion Models: An Information-Theoretic Perspective

University of California, Riverside, CA, US(加州大学河滨分校)

Shaorong Zhang, Longxuan Yu, Rob Brekelmans, Luhan Tang, Salman Asif, Greg Ver Steeg

AI总结:

本文从信息论视角探讨了掩码扩散模型中生成顺序与并行解码的理论机制,揭示了顺序敏感性与并行化偏差的内在关系,并通过实验验证了理论框架的有效性。

摘要:

掩码扩散模型(MDMs)通过牺牲顺序确定性显著加速推理过程。然而,生成顺序的理论机制以及并行化固有的风险仍缺乏深入研究。在本工作中,我们提供了一个统一的信息论框架,用于解耦和分析两个根本性的失败来源:顺序敏感性和并行化偏差。我们的分析得出三个关键见解:(1)在模型误差增加时,优先处理低熵token的易先解码的优势被放大;(2)因子化并行解码引入内在采样误差,可能导致任意大的反向KL散度,捕捉到标准前向KL度量所忽视的“不一致性”失败;(3)虽然验证可以消除采样误差,但会带来由块内总相关性决定的指数级成本。相反,诸如重掩码等启发法虽然计算高效,但无法保证分布正确性。在受控的块HMM和大规模MDMs(LLaDA)上进行的算术推理实验验证了我们的理论框架。

英文摘要:

Masked Diffusion Models (MDMs) significantly accelerate inference by trading off sequential determinism. However, the theoretical mechanisms governing generation order and the risks inherent in parallelization remain under-explored. In this work, we provide a unified information-theoretic framework to decouple and analyze two fundamental sources of failure: order sensitivity and parallelization bias. Our analysis yields three key insights: (1) The benefits of Easy-First decoding (prioritizing low-entropy tokens) are magnified as model error increases; (2) factorized parallel decoding introduces intrinsic sampling errors that can lead to arbitrary large Reverse KL divergence, capturing "incoherence" failures that standard Forward KL metrics overlook; and (3) while verification can eliminate sampling error, it incurs an exponential cost governed by the total correlation within a block. Conversely, heuristics like remasking, though computationally efficient, cannot guarantee distributional correctness. Experiments on a controlled Block-HMM and large-scale MDMs (LLaDA) for arithmetic reasoning validate our theoretical framework.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑