Prefix-Adaptive Block Diffusion for Efficient Document Recognition

TL;DR

The paper introduces PA-BDM, a novel diffusion model for document recognition that enhances efficiency and accuracy by adaptively managing prefix commitment and denoising strategies.

Contribution

It proposes a prefix-adaptive diffusion approach with confidence-gated loss and progressive prefix commitment, improving parallel decoding and recognition performance.

Findings

PA-BDM achieves higher recognition scores on benchmarks.

Inference throughput improves by 71.6% over previous models.

Replaces intra-block bidirectional denoising with causal denoising.

Abstract

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.