PulseAugur
EN
LIVE 08:47:42
中文(ZH) 百度OCR模型Unlimited OCR在HuggingFace、GitHub四榜登顶

Baidu's Unlimited OCR model tops leaderboards and sees rapid adoption

Baidu has released and open-sourced its end-to-end OCR model, Unlimited OCR. The model quickly achieved top rankings on GitHub and Hugging Face, demonstrating strong performance in long document parsing. Unlimited OCR achieved a new record on the OmniDocBench v1.6 benchmark with a 93.92% comprehensive score, and has seen rapid adoption and local deployment. AI

IMPACT Sets a new benchmark for OCR performance and demonstrates the rapid spread of open-source AI models for document processing.

RANK_REASON Open-source release of a new OCR model by a major tech company (Baidu) with strong benchmark performance. [lever_c_demoted from frontier_release: ic=2 ai=1.0]

Read on 36氪 (36Kr) →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

Baidu's Unlimited OCR model tops leaderboards and sees rapid adoption

COVERAGE [2]

  1. 36氪 (36Kr) TIER_1 中文(ZH) ·

    Baidu OCR Model Unlimited OCR Tops Four Charts on HuggingFace and GitHub

    36氪获悉,近日,百度正式发布并开源端到端OCR模型Unlimited OCR。模型发布次日即登顶GitHub Daily Trending榜、Python榜,在HuggingFace全球模型总趋势榜和多模态模型趋势榜也均排名第一。Unlimited OCR面向长文档解析场景打造,总参数规模3B、推理时激活参数仅约570M。公开评测结果显示,Unlimited OCR在OmniDocBench v1.6基准测试中取得 93.92% 综合成绩,刷新端到端OCR最新纪录。

  2. Mastodon — sigmoid.social TIER_1 한국어(KO) · [email protected] ·

    thehype. (@thehypedotnews) Baidu's unlimited-ocr ranked 2nd trending on Hugging Face with 70k+ downloads, and its OCR performance was tested by scanning actual newspaper/stock pages locally. Publicly released OCR models are spreading rapidly, with local deployment

    thehype. (@thehypedotnews) Baidu의 unlimited-ocr가 Hugging Face에서 2위 트렌딩과 7만+ 다운로드를 기록했고, 로컬에서 실행해 실제 신문/증권 페이지 스캔으로 OCR 성능을 테스트했다. 공개된 OCR 모델이 빠르게 확산 중이며, 로컬 배포 가능성과 문서 인식 워크플로우에서의 실사용성을 가늠해볼 만한 소식이다. https:// x.com/thehypedotnews/status/20 70537465706799120 # ocr # huggingface # …