Prefix-Adaptive Block Diffusion for Efficient Document Recognition
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918506311712768 |
|---|---|
| author | Chai, Mingxu Shen, Ziyu Liu, Chenyu Zhang, Kaidi Zhang, Jiazheng Zhu, Dingwei Xi, Zhiheng Chen, Ruoyu Long, Jun Kang, Jihua Gui, Tao Zhang, Qi |
| author_facet | Chai, Mingxu Shen, Ziyu Liu, Chenyu Zhang, Kaidi Zhang, Jiazheng Zhu, Dingwei Xi, Zhiheng Chen, Ruoyu Long, Jun Kang, Jihua Gui, Tao Zhang, Qi |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_16861 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Prefix-Adaptive Block Diffusion for Efficient Document Recognition Chai, Mingxu Shen, Ziyu Liu, Chenyu Zhang, Kaidi Zhang, Jiazheng Zhu, Dingwei Xi, Zhiheng Chen, Ruoyu Long, Jun Kang, Jihua Gui, Tao Zhang, Qi Computer Vision and Pattern Recognition Artificial Intelligence 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. |
| title | Prefix-Adaptive Block Diffusion for Efficient Document Recognition |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2605.16861 |