mlsanityData to 14 Oct 2026

PaDoc: Layout-Grounded Parallel Decoding for Document Parsing

arXiv 2608.06146 ↗First posted 6 Aug 2026cs.AIcited by 1

On the map: Efficient LLM inference and serving → Visual token pruning in multimodal models

Abstract (arXiv)

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

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Visual token pruning in multimodal models, one of 12 directions in Efficient LLM inference and serving. Builds on 27 of its 32 references that are in the map. First posted on arXiv 6 Aug 2026.

Builds on

27 of 32 references.

References · 27

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,653
  4. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 843
  5. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertaintycited by 578
  6. 2022Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understandingcited by 366
  7. 2023Nougat: Neural Optical Understanding for Academic Documentscited by 245
  8. 2025MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsingcited by 116
  9. 2023Skeleton-of-Thought: Prompting LLMs for Efficient Parallel Generationcited by 105
  10. 2024DocLayout-YOLO: Enhancing Document Layout Analysis through Diverse Synthetic Data and Global-to-Local Adaptive Perceptioncited by 82
  11. 2025Dolphin: Document Image Parsing via Heterogeneous Anchor Promptingcited by 79
  12. 2023SPTS v2: Single-Point Scene Text Spottingcited by 66
15 more
  1. 2026PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsingcited by 57
  2. 2025SmolDocling: An ultra-compact vision-language model for end-to-end multi-modal document conversioncited by 53
  3. 2025Hogwild! Inference: Parallel LLM Generation via Concurrent Attentioncited by 32
  4. 2025Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decodingcited by 28
  5. 2026MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scalecited by 27
  6. 2025PP-DocLayout: A Unified Document Layout Detection Model to Accelerate Large-Scale Data Constructioncited by 26
  7. 2024APAR: LLMs Can Do Auto-Parallel Auto-Regressive Decodingcited by 24
  8. 2026Qianfan-OCR: A Unified End-to-End Model for Document Intelligencecited by 21
  9. 2024DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inferencecited by 19
  10. 2026Multimodal OCR: Parse Anything from Documentscited by 14
  11. 2026HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Bettercited by 14
  12. 2025ASPD: Unlocking Adaptive Serial-Parallel Decoding by Exploring Intrinsic Parallelism in LLMscited by 11
  13. 2025SPRINT: Enabling Interleaved Planning and Parallelized Execution in Reasoning Modelscited by 10
  14. 2026HSD: Training-Free Acceleration for Document Parsing Vision-Language Models with Hierarchical Speculative Decodingcited by 6
  15. 2026MonkeyOCRv2: A Visual-Text Foundation Model for Document AIcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026WeVisDoc: From Coverage to Capability for Robust End-to-End Document Parsingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026HPD-Parsing: Hierarchical Parallel Document Parsing
  2. 2026Efficient Document Parsing via Parallel Token Prediction
  3. 2026Towards Real-World Document Parsing via Realistic Scene Synthesis and Document-Aware Training
  4. 2026Structured Layout Priors for Robust Out-of-Distribution Visual Document Understanding
  5. 2026Dolphin-v2: Universal Document Parsing via Scalable Anchor Prompting
  6. 2025Infinity Parser: Layout Aware Reinforcement Learning for Scanned Document Parsing
  7. 2026Read, Parse, Describe: Unified Document Parsing with Visual Element Description Generation
  8. 2026Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing