B3 方法让扩散语言模型并行去噪两个块，提速 11.5%-14.5%
Blurring the Block Boundaries: Overlapping Block Denoising for Efficient Diffusion Language Models
做低延迟生成的朋友可以看看这篇：DLM 的块扩散原本一块卡住全卡住，B3 让两个块在一次前向传播里并行去噪，不用训练不用调参，10 个数据集上白拿 11.5%-14.5% 加速且质量不掉。
一篇新论文提出 Blurring the Block Boundaries（B3），针对自回归之外的新兴候选 Diffusion Language Models（DLM）的推理瓶颈：块扩散（block diffusion）中每个块要等最慢的 token 完成才能进入下一块。B3 在共享前向传播中并行精炼两个块，并动态校准，只有当模型准备好时才打开下一个块。
该方法无需训练、无需调超参数，可以直接嵌入任何带自编辑（self-editing）能力的块扩散模型。在覆盖多种任务和难度级别的 10 个数据集上测试，B3 在生成质量持平的情况下带来 11.5%-14.5% 的加速，推进了 DLM 推理在延迟-质量权衡上的 Pareto 前沿。
对于关注低延迟文本生成的开发者，B3 提供了一条不改权重就能白捡速度的路径，其“重叠块”思路也可能被后续块扩散模型直接采纳为默认推理策略。
Blurring the Block Boundaries: Overlapping Block Denoising for Efficient Diffusion Language Models
Diffusion Language Models (DLMs) have emerged as effective alternatives to autoregressive models for latency-critical tasks, enabling faster token generation. Recent advances employ block diffusion with self-editing capabilities, where each block refines tokens iteratively before advancing to the next one. Yet each block is only as fast as its slowest token, and the next block cannot start before it. We suggest Blurring the Block Boundaries (B3), a novel method with which two blocks are refined in parallel within a shared forward pass. B3 calibrates itself on the fly, opening the next block only once the model is ready to start generating it. It is a training-free and hyperparameter-free acceleration method which seamlessly integrates within any self-editing block diffusion model. Evaluated on state-of-the-art DLMs across 10 diverse datasets spanning varied tasks and difficulty levels, B3 results in on par generation quality and 11.5-14.5% speedup compared to the base model, pushing the Pareto frontier for latency-quality trade-offs in DLM inference.
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Diffusion Language Models, block diffusion, 推理加速, 开源模型.
Source attribution: curated from arXiv cs.AI (@Amir Mann, Jonathan Fhima, Shaked Brody, Or Avitan, Asif Zamir, Shahar Tsiper, Niv Nayman).