github opendatalab/MinerU mineru-4.1.4-released
MinerU 4.1.4

3 hours ago

中文

  • Standard / Advanced 的 PDF 文本模式使用 VLM 识别表格外的行内公式切片,保留混合文本与公式;单个切片识别失败时局部回退 MFR。
  • 改善公式定界符和文本转义处理,避免普通文本中的美元符号被误当成公式或携带多余转义字符。
  • 将 DocVortex 最低版本提升至 0.6.11,并在中英文 README 中补充解析档位指南入口。

English

  • Use VLM text recognition for inline formula crops outside tables in Standard / Advanced PDF text mode, preserving mixed text and math with per-crop MFR fallback.
  • Improve math-delimiter and text-escape handling so ordinary dollar signs remain text without redundant escape characters.
  • Require DocVortex 0.6.11 or later within the 0.6 series and link parsing-tier guides from both READMEs.

Related PR: #5706

Full Changelog: mineru-4.1.3-released...mineru-4.1.4-released

Don't miss a new MinerU release

NewReleases is sending notifications on new releases.