Issues

• [v2][GB10 / SM121] SGLang server degenerates from the first tokens (bbox coordinate loops); Transformers output on the same page is correct

  baidu/Unlimited-OCR#114 · Filter by author SergiFuster opened on Oct 28, 2026
  2

• critical security issue

  baidu/Unlimited-OCR#113 · Filter by author Zero2oneZ opened on Oct 26, 2026

• ocr

  baidu/Unlimited-OCR#112 · Filter by author sajidsuhile opened on Oct 26, 2026

• Word-level bounding boxes

  baidu/Unlimited-OCR#108 · Filter by author mkrzywda opened on Oct 18, 2026

• Inconsistent Page Separator Extraction During Multi-Page OCR Inference

  baidu/Unlimited-OCR#107 · Filter by author Navani001 opened on Oct 18, 2026

• infer.py requests honor proxy env vars — loopback health checks and inference calls fail on proxy-configured hosts (README's client example already sets trust_env=False)

  baidu/Unlimited-OCR#105 · Filter by author rahul05ranjan opened on Oct 15, 2026
  1

• infer.py: input validation runs only after the SGLang server is launched (missing --pdf/--image_dir costs a full model load)

  baidu/Unlimited-OCR#104 · Filter by author rahul05ranjan opened on Oct 15, 2026

• infer.py always exits 0 — even when every request fails or no inputs are found

  baidu/Unlimited-OCR#103 · Filter by author rahul05ranjan opened on Oct 15, 2026
  1

• infer.py: --image_dir output filenames collide — sub/doc.png and sub__doc.png both write sub__doc.md

  baidu/Unlimited-OCR#102 · Filter by author rahul05ranjan opened on Oct 15, 2026
  1

• R&D feedback: model is 8-19x slower than Tesseract per page, even on GPU - too heavy for our lightweight ingestion pipeline

  baidu/Unlimited-OCR#100 · Filter by author mohammad-ibrahim-enosis opened on Sep 28, 2026

• Docs request: community-tested Intel Arc A770/Vulkan single-page starter path (GGUF)

  baidu/Unlimited-OCR#99 · Filter by author twinforcesl opened on Sep 25, 2026

• [Feature Request] Add Agent skill/tool integration interface for OCR capability

  baidu/Unlimited-OCR#97 · Filter by author jexbow opened on Aug 31, 2026
  3

• ArabicOCR

  baidu/Unlimited-OCR#95 · Filter by author BizMapper opened on Aug 12, 2026

• Is the source code for the Hugging Face Spaces web demo available?

  baidu/Unlimited-OCR#94 · Filter by author GaryHu9527 opened on Aug 12, 2026

• 模型推理部署

  baidu/Unlimited-OCR#93 · Filter by author xh19980728-creator opened on Aug 10, 2026

• Hebrew text: model outputs a hallucinated token while Latin text on the same image is read correctly — is Hebrew supported?

  baidu/Unlimited-OCR#90 · Filter by author aseel1 opened on Aug 4, 2026

• Enhancement / Discussion: Benchmark and Support for Arabic Language Parsing via arafix

  baidu/Unlimited-OCR#89 · Filter by author bio-colab opened on Aug 3, 2026

• dokOSR

  baidu/Unlimited-OCR#88 · Filter by author Armany7878 opened on Aug 2, 2026

• Herramientas

  baidu/Unlimited-OCR#87 · Filter by author jimmiord3005-cpu opened on Jul 31, 2026
  1

• Patent's ST.25 listings become tables and digit-changing generation loops in vLLM

  baidu/Unlimited-OCR#84 · Filter by author Linnas opened on Jul 26, 2026

• Output texts keeps on repeating infinitely [Tried with llama.cpp server running on ubuntu 22.04]

  baidu/Unlimited-OCR#82 · Filter by author masifamu opened on Jul 25, 2026
  5

• Non-CUDA inference: CPU works after removing hardcoded .cuda(); MPS (Apple Silicon) crashes mid-generation

  baidu/Unlimited-OCR#81 · Filter by author GregoriusAlvin1 opened on Jul 23, 2026
  2

• Rotated text parsing drops multiple text lines and generates a spurious line during OCR

  baidu/Unlimited-OCR#79 · Filter by author ezManish opened on Jul 23, 2026

• Improve error handling for unsupported input formats in model.infer()

  baidu/Unlimited-OCR#78 · Filter by author ezManish opened on Jul 23, 2026

• Total number of attention heads (10) must be divisible by tensor parallel size (4)

  baidu/Unlimited-OCR#76 · Filter by author zgsxwsdxg opened on Jul 23, 2026
  1