这篇论文解决了扩散语言模型在训练和推理时配置不一致的痛点，用ABD方法在多种配置下训练一个模型就能通用，不用再为不同解码策略分别训练了。
扩散语言模型(DLM)通常在固定上下文结构下训练，限制去噪到预定token子集，这导致训练与推理不匹配（模型需在任意配置下运行，训练网格外表现下降）。提出的Adaptive Block Diffusion (ABD)通过优化前缀窗口配置分布上的去噪风险来消除此不匹配，将配置视为随机变量，无需架构修改即可在全配置空间训练单一模型。实验表明ABD在不同解码尺度下保持结构一致性，避免离网格崩溃，恢复块大小与困惑度之间的单调关系，并在目标尺度上匹配或超越固定块专家。
Adaptive Block Diffusion: Resolving Training-Inference Mismatch in Diffusion Language Models
Diffusion Language Models (DLMs) are typically trained under fixed context structures, restricting denoising to predetermined token subsets. This creates a mismatch between training and inference, where models must operate over arbitrary configurations, leading to degradation off the training grid. We propose Adaptive Block Diffusion (ABD), which resolves this mismatch by optimizing denoising risk over a distribution of prefix-window configurations. By treating the configuration as a stochastic variable, ABD trains a single model over the full configuration space without architectural changes. We show that generalization across decoding strategies is governed by the support of the training distribution, and that ABD guarantees denoising optimality for any inference policy whose configurations are covered during training. Empirically, ABD exhibits structural invariance across decoding scales, avoiding off-grid collapse and recovering a monotonic relationship between block size and perplexity, while matching or outperforming fixed-block specialists at their target scales.