v3.5.0
·
6 commits
to release/3.5
since this release
2026.4.17 v3.5.0 released
- Supports switching the underlying inference engine, with options such as PaddlePaddle framework and Transformers.
- Pipelines such as PaddleOCR-VL series and PP-StructureV3 support returning DOCX documents (PDF/image to DOCX) in parsing results.
2026.4.17 v3.5.0 发布
- 支持更换底层推理引擎，可选飞桨框架、Transformers 等。
- PaddleOCR-VL 系列、PP-StructureV3 等产线支持在解析结果中返回 DOCX 文档（PDF/Image to DOCX）。
Full Changelog: v3.4.0...v3.5.0