Baidu releases Unlimited OCR 2, an MIT-licensed long-document parser
Baidu's Hugging Face model card describes Unlimited OCR 2, a vision-language document parser released under the MIT license. It introduces Sparse Reference Sliding Window Attention (SR-SWA), which compresses attention over long sequences into a fixed-size set of working references, e.g. 512 tokens. The card states this keeps per-step decoding latency and memory footprint nearly constant regardless of document length. It supports multi-page inference via Transformers and recommends SGLang for serving, with configuration guidance for memory tuning.
Open original source huggingface.coOne primary source; other links are context, not independent corroboration.
- Primary sourceX discovery · automated brief
- Discovery linkX
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
- Original repository+40
- Feed text or abstract+15
- No editor link review0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
Baidu's Hugging Face model card describes Unlimited OCR 2, a vision-language document parser released under the MIT license. It introduces Sparse Reference Sliding Window Attention (SR-SWA), which compresses attention over long sequences into a fixed-size set of working references, e.g. 512 tokens. The card states this keeps per-step decoding latency and memory footprint nearly constant regardless of document length. It supports multi-page inference via Transformers and recommends SGLang for serving, with configuration guidance for memory tuning.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: huggingface.co. Brief basis: Hugging Face model card. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Source claims not independently checked; the card's performance claims rest on its own leaderboard and benchmark results, and the card itself notes that smaller page windows degrade cross-page merging.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- SR-SWA compresses text-visual attention to a fixed-size set of working references, e.g. 512 tokens.
- The card says per-step decoding latency and memory footprint stay nearly constant regardless of document length.
- The model parses all pages of a document in one request via infer_multi, using cross-page context.
- Serving is recommended through SGLang, with page-window scheduling for very long documents.
Source claims not independently checked; the card's performance claims rest on its own leaderboard and benchmark results, and the card itself notes that smaller page windows degrade cross-page merging.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
These are Grok Bot discovery notes. AINewsIQ prepared a brief from available source material. The bot's rationale, linked posts, authors, and stances have not been independently verified.
Baidu released Unlimited OCR 2 on Hugging Face (MIT). Its SR-SWA attention shrinks text-visual attention to a fixed set of working references, so per-step decoding latency and memory stay nearly constant as documents grow. Baidu claims state-of-the-art results on MPDocBench and the 100-page MarathonDocBench. Benchmarks are self-reported.
Source claims not independently checked; the card's performance claims rest on its own leaderboard and benchmark results, and the card itself notes that smaller page windows degrade cross-page merging.
Engagement counts, when shown, are observed snapshots. They do not establish accuracy or corroboration.
- huggingface post ↗ · 124 likes, 513 downloads · observed 10/27/2026, 10:09:04 AM
- x post ↗ · 523 likes, 62 reposts, 32 replies, 39 quotes · observed 10/27/2026, 10:07:53 AM
- x post ↗ · 10 likes, 2 reposts, 2 replies, 0 quotes · observed 10/27/2026, 10:07:53 AM
- reddit post ↗
- Related post ↗
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from Hugging Face model card.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
Long-document OCR is limited by quadratic attention cost and memory growth. An open, MIT-licensed model claiming bounded per-step cost on 100+ page documents could make large-scale document parsing cheaper to run, if the claims hold.
Limits and open questions
Source claims not independently checked; the card's performance claims rest on its own leaderboard and benchmark results, and the card itself notes that smaller page windows degrade cross-page merging.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.