Hao Yu, Jiabo Zhan, Kang Liu, Linnan Zhao, Dongxu Yue, Rui Chen, Jinglin Wang, Chong Sun, Chen Li, Jing LYU, Chun Yuan
3 min
Abstract
End-to-end document parsers provide a unified interface, but serialize page layouts and regional contents into one autoregressive sequence. This formulation forces independent regions onto a decoding path whose length grows with the total content, whereas crop-based two-stage parsers expose region-level parallelism at the cost of repeated visual prefills and fragmented page context. To retain full-page context while removing dependencies, we propose PaDoc, a layout-grounded parser that treats the predicted layout as a branching structure over a shared page representation. Under a region-sufficiency assumption, we derive a prefix-conditioned factorization in which the layout stream and regional content branches advance concurrently, reducing the decoding depth to the longest layout-content path. We realize this factorization within a single MLLM: packed variable-length ancestor attention preserves the visibility under standard next-token training, while masked parallel decoding creates branches that the evaluated vLLM backend serves as concurrent requests with cache-resident shared-prefix reuse. On OmniDocBench Full, PaDoc attains an Overall layout F1 of 91.1 and, among end-to-end parsers, a top-tier Overall score of 94.24 together with the best Text Edit (0.038) and Formula CDM (95.59). On a 384-page subset and one A800 GPU, it is the fastest end-to-end parser at five concurrency levels, improving valid-page throughput by 67.4-118% and reducing P95 latency by 39.2-54.9% relative to a same-backbone Sequential SFT baseline. Code is available at https://github.com/Longin-Yu/Padoc
Sam: So the branch always knows both where it sits on the page and where it sits in the document's hierarchy.
Alex: Right. That combination — shared visual context plus hierarchical memory — is what allows the parallel processing to stay coherent.
Sam: Did the paper show a meaningful difference in speed?
Alex: The results are notable. Across their tests, the system more than doubled throughput in some cases, and cut the time needed to finish processing a document by more than half. And critically, it achieves this without adding new components to the underlying model. It works within the existing architecture.
Sam: So the gain isn't coming from more computing power or a bigger model. It's coming from a smarter way of organizing the work.
Alex: That's the key finding. The same model, doing the same job, but structured differently — and the result is a meaningful improvement in efficiency. It suggests that for complex documents, how you sequence the problem matters just as much as how much processing power you throw at it. Thanks for listening to ResearchPod.