HPD-Parsing: Hierarchical Parallel Document Parsing
Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.
A New Paradigm for Document Parsing from the PaddleOCR Team</p>\n","updatedAt":"2026-07-22T04:20:45.479Z","author":{"_id":"65a5231a087d8a2e9cc2414b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65a5231a087d8a2e9cc2414b/wj0l5R5LmBUG-E8XTMdBM.jpeg","fullname":"cuicheng","name":"ChengCui","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":40,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7702298164367676},"editors":["ChengCui"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/65a5231a087d8a2e9cc2414b/wj0l5R5LmBUG-E8XTMdBM.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.18839","authors":[{"_id":"6a6036b77e7f152167e47132","user":{"_id":"630ec94805b5c5280e3bf194","avatarUrl":"/avatars/3bd98d7e9bfc402f198e2ada704cefe7.svg","isPro":false,"fullname":"WEI","user":"WEISHU","type":"user","name":"WEISHU"},"name":"Shu Wei","status":"claimed_verified","statusLastChangedAt":"2026-07-22T07:40:47.143Z","hidden":false},{"_id":"6a6036b77e7f152167e47133","name":"Jingjing Wu","hidden":false},{"_id":"6a6036b77e7f152167e47134","name":"Lingshu Zhang","hidden":false},{"_id":"6a6036b77e7f152167e47135","name":"Qunyi Xie","hidden":false},{"_id":"6a6036b77e7f152167e47136","name":"Hao Zou","hidden":false},{"_id":"6a6036b77e7f152167e47137","name":"Le Xiang","hidden":false},{"_id":"6a6036b77e7f152167e47138","name":"Xu Fan","hidden":false},{"_id":"6a6036b77e7f152167e47139","name":"Yangliu Xu","hidden":false},{"_id":"6a6036b77e7f152167e4713a","user":{"_id":"67d96e68939c3823ab2e06a5","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/FJx_Nf_5aD1w1s0RqbyGn.png","isPro":false,"fullname":"Manhui Lin","user":"gggdddfff","type":"user","name":"gggdddfff"},"name":"Manhui Lin","status":"claimed_verified","statusLastChangedAt":"2026-07-22T07:39:29.335Z","hidden":false},{"_id":"6a6036b77e7f152167e4713b","name":"Xiaolong Ma","hidden":false},{"_id":"6a6036b77e7f152167e4713c","user":{"_id":"65a5231a087d8a2e9cc2414b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65a5231a087d8a2e9cc2414b/wj0l5R5LmBUG-E8XTMdBM.jpeg","isPro":false,"fullname":"cuicheng","user":"ChengCui","type":"user","name":"ChengCui"},"name":"Cheng Cui","status":"claimed_verified","statusLastChangedAt":"2026-07-22T07:39:29.328Z","hidden":false},{"_id":"6a6036b77e7f152167e4713d","name":"Tengyu Du","hidden":false},{"_id":"6a6036b77e7f152167e4713e","name":"YY","hidden":false}],"publishedAt":"2026-07-21T00:00:00.000Z","submittedOnDailyAt":"2026-07-22T00:00:00.000Z","title":"HPD-Parsing: Hierarchical Parallel Document Parsing","submittedOnDailyBy":{"_id":"65a5231a087d8a2e9cc2414b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65a5231a087d8a2e9cc2414b/wj0l5R5LmBUG-E8XTMdBM.jpeg","isPro":false,"fullname":"cuicheng","user":"ChengCui","type":"user","name":"ChengCui"},"summary":"Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jointly but generate its output through a single token-by-token autoregressive trajectory, creating a sequential bottleneck that grows with document length. Such full-page sequential generation overlooks a key property of document parsing: layout must be analyzed globally, whereas block content can be parsed in parallel. Based on this observation, we introduce HPD-Parsing, which replaces full-page autoregressive generation with a Hierarchical Parallel Decoding paradigm. A main layout branch organizes the overall document structure and dynamically assigns block-level content decoding to concurrent branches, while progressive multi-token prediction (P-MTP) further reduces the decoding steps within each branch. Experiments on public benchmarks show that HPD-Parsing achieves 4,752 tokens per second, delivering 2.62times the throughput of the fastest existing document parsing model and 3.06times that of the vanilla autoregressive baseline, while maintaining competitive parsing accuracy. These results establish hierarchical parallel decoding as an effective alternative to full-page autoregressive generation, opening a new direction for efficient unified document parsing.","upvotes":5,"discussionId":"6a6036b77e7f152167e4713f","organization":{"_id":"62067d5d3906f102bc9658bd","name":"PaddlePaddle","fullname":"PaddlePaddle","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1654942635336-5f3ff69679c1ba4c353d0c5a.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"65a5231a087d8a2e9cc2414b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65a5231a087d8a2e9cc2414b/wj0l5R5LmBUG-E8XTMdBM.jpeg","isPro":false,"fullname":"cuicheng","user":"ChengCui","type":"user"},{"_id":"630ec94805b5c5280e3bf194","avatarUrl":"/avatars/3bd98d7e9bfc402f198e2ada704cefe7.svg","isPro":false,"fullname":"WEI","user":"WEISHU","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"6780890b816943dee855d100","avatarUrl":"/avatars/cd73a57d8078b6ede0a7f0aee04b28d8.svg","isPro":false,"fullname":"Wujingjing","user":"wjj1999","type":"user"},{"_id":"65e48259a4e46e644ebf4f6a","avatarUrl":"/avatars/4950c9e0b1bedd357fd6fa606665d4a5.svg","isPro":false,"fullname":"Johny_XIE","user":"johnyxie","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"62067d5d3906f102bc9658bd","name":"PaddlePaddle","fullname":"PaddlePaddle","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1654942635336-5f3ff69679c1ba4c353d0c5a.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.18839.md","query":{}}">
Papers
arxiv:2607.18839
HPD-Parsing: Hierarchical Parallel Document Parsing
Abstract
Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jointly but generate its output through a single token-by-token autoregressive trajectory, creating a sequential bottleneck that grows with document length. Such full-page sequential generation overlooks a key property of document parsing: layout must be analyzed globally, whereas block content can be parsed in parallel. Based on this observation, we introduce HPD-Parsing, which replaces full-page autoregressive generation with a Hierarchical Parallel Decoding paradigm. A main layout branch organizes the overall document structure and dynamically assigns block-level content decoding to concurrent branches, while progressive multi-token prediction (P-MTP) further reduces the decoding steps within each branch. Experiments on public benchmarks show that HPD-Parsing achieves 4,752 tokens per second, delivering 2.62times the throughput of the fastest existing document parsing model and 3.06times that of the vanilla autoregressive baseline, while maintaining competitive parsing accuracy. These results establish hierarchical parallel decoding as an effective alternative to full-page autoregressive generation, opening a new direction for efficient unified document parsing.
Community
A New Paradigm for Document Parsing from the PaddleOCR Team
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Get this paper in your agent:
hf papers read 2607.18839 curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper
Datasets citing this paper
No dataset linking this paper
Cite arxiv.org/abs/2607.18839 in a dataset README.md to link it from this page.
Spaces citing this paper
Collections including this paper
No Collection including this paper
Add this paper to a collection to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.