提出新型块扩散模型，实现高效文档识别与并行推理。
Prefix-Adaptive Block Diffusion for Efficient Document Recognition
- 用前缀自回归替代块内双向去噪，提升推理效率。
- 在多个基准上优于2.5B MinerU-Diffusion，推理吞吐提升71.6%。
- 适合需要长序列生成与高吞吐的文档结构识别场景。
块扩散模型（BDMs）支持并行生成、灵活长度输出和键值缓存，适用于高效文档解析。然而现有BDM将去噪与缓存绑定于固定块边界：块内去噪时并行性下降，且只有整块生成后才能缓存。此外，块内双向去噪与块间自回归冲突，导致信息流不一致，影响结构敏感识别。我们提出前缀自适应块扩散模型（PA-BDM），以从前缀到后缀的因果去噪取代块内双向去噪，并将块大小视为最大候选范围而非固定承诺单元。PA-BDM采用置信度门控结构损失（CSL）在训练中构建低熵前缀，以支持更长延续。推理时，渐进式前缀承诺（PPC）动态将最长可靠前缀写入键值缓存，并从更新后的前缀重置下一候选范围，每步恢复大并行解码空间。实验表明，3B PA-BDM在多个基准上表现更优，推理吞吐比2.5B MinerU-Diffusion提升71.6%。
原文摘要 · Abstract (English)
Block Diffusion Models (BDMs) support parallel generation, flexible-length output, and KV caching, making them promising for efficient document parsing. However, existing BDMs bind denoising and cache commitment to fixed block boundaries: parallelism shrinks during intra-block denoising, while generated tokens cannot be cached until the whole block is completed. Moreover, intra-block bidirectional denoising conflicts with inter-block autoregression, creating inconsistent information flow that can challenge structure-sensitive recognition. We propose the Prefix-Adaptive Block Diffusion Model (PA-BDM), which replaces intra-block bidirectional denoising with causal denoising from prefix to suffix and treats the block size as a maximum candidate range rather than a fixed commitment unit. PA-BDM uses Confidence-gated Structural Loss (CSL) to build low-entropy prefixes before extending training to longer continuations. During inference, Progressive Prefix Commitment (PPC) then dynamically commits the longest reliable prefix into the KV cache and resets the next candidate range from the updated prefix, restoring a large parallel decoding space at each step. Experiments show that the 3B PA-BDM achieves higher recognition scores on several benchmarks and improves inference throughput by 71.6\% over the 2.5B MinerU-Diffusion.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。