PaddleOCR-VL Collection Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model • 3 items • Updated Oct 17 • 21