基于置信度的解码在扩散语言模型中具有可证明的效率
Confidence-Based Decoding is Provably Efficient for Diffusion Language Models
AI总结：
本文提出了一种基于熵和的解码策略，证明其在KL散度下具有ε精度，并在迭代次数上达到O(H(X0)/ε)的效率，适用于低熵数据分布。
摘要：
扩散语言模型（DLMs）作为一种有前景的替代自回归（AR）模型，允许灵活的生成顺序和并行生成多个token。然而，这种灵活性引入了在AR模型中不存在的挑战：解码策略——决定每次迭代中生成的token顺序和数量——对采样效率有关键影响。在实践中探索的解码策略中，基于置信度的方法通过根据预测置信度适应性地选择哪些和多少token进行解掩码，表现出强大的经验性能。尽管取得了成功，我们对基于置信度的解码的理论理解仍然有限。在本文中，我们开发了首个针对DLMs中基于置信度的解码的理论分析框架。我们专注于一种基于熵和的策略，在每次迭代中继续解掩码token，直到累积熵超过阈值，并证明其在KL散度下达到ε精度的采样，其期望迭代次数为O(H(X0)/ε)，其中H(X0)表示目标数据分布的熵。值得注意的是，当数据分布相对于序列长度具有低熵时，该策略能显著提高采样速度，同时自动适应数据的内在复杂性，而无需先验知识或超参数调整。总体而言，我们的结果为基于置信度的解码提供了理论基础，并可能指导设计更高效的DLMs解码策略。
英文摘要：
Diffusion language models (DLMs) have emerged as a promising alternative to autoregressive (AR) models for language modeling, allowing flexible generation order and parallel generation of multiple tokens. However, this flexibility introduces a challenge absent in AR models: the \emph{decoding strategy} -- which determines the order and number of tokens generated at each iteration -- critically affects sampling efficiency. Among decoding strategies explored in practice, confidence-based methods, which adaptively select which and how many tokens to unmask based on prediction confidence, have shown strong empirical performance. Despite this success, our theoretical understanding of confidence-based decoding remains limited. In this work, we develop the first theoretical analysis framework for confidence-based decoding in DLMs. We focus on an entropy sum-based strategy that continues unmasking tokens within each iteration until the cumulative entropy exceeds a threshold, and show that it achieves $\varepsilon$-accurate sampling in KL divergence with an expected number of iterations $\widetilde O(H(X_0)/\varepsilon)$, where $H(X_0)$ denotes the entropy of the target data distribution. Notably, this strategy yields substantial sampling acceleration when the data distribution has low entropy relative to the sequence length, while automatically adapting to the intrinsic complexity of data without requiring prior knowledge or hyperparameter tuning. Overall, our results provide a theoretical foundation for confidence-based decoding and may inform the design of more efficient decoding strategies for DLMs.
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。
