Baidu has released and open-sourced Wenxin Unlimited OCR 2, a new OCR model capable of parsing documents of any length. This advancement pushes the boundaries of OCR technology into a new era, enabling the analysis of entire books or extensive reports with high accuracy and efficiency. The model achieves this through a novel Sparse Reference Sliding Window Attention (SR-SWA) mechanism, which significantly speeds up processing for long documents by focusing attention on relevant sections rather than the entire text. AI
IMPACT Enables analysis of entire books and extensive reports, potentially accelerating research and document processing workflows.
RANK_REASON New model release from a major AI lab (Baidu) with significant technical innovation and benchmark performance claims. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Baidu
- GitHub
- Hugging Face
- OCR 3.0
- Sparse Reference Sliding Window Attention
- SR-SWA
- Unlimited OCR 2
- Yann LeCun
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →