BHL OCR Leaderboard: likes over time

finebooks/bhl-ocr-leaderboard is a static Space on the Hugging Face Hub. It has 17 likes, 0 in the last 7 days and 3 in the last 30, and ranks #6,099 among Spaces by likes.

OCR & VLM character error rates on 18th–19th-c. book pages

What it uses

• Qwen/Qwen3.5-9B (model)
• deepseek-ai/DeepSeek-OCR (model)
• zai-org/GLM-OCR (model)
• baidu/Unlimited-OCR (model)
• deepseek-ai/DeepSeek-OCR-2 (model)
• ATH-MaaS/OvisOCR2 (model)
• lightonai/LightOnOCR-2-1B (model)
• baidu/Qianfan-OCR (model)
• allenai/olmOCR-2-7B-1025-FP8 (model)
• PaddlePaddle/PaddleOCR-VL-1.6 (model)
• ds4sd/SmolDocling-256M-preview (model)
• rednote-hilab/dots.mocr (model)
• rednote-hilab/dots.ocr (model)
• finebooks/bhl-impact-gt (dataset)

Open finebooks/bhl-ocr-leaderboard on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.