Test dataset Version

This version includes all features in the test dataset.

Training dataset Version

Merge pull request #1 from UniModal4Reasoning/dev

Training dataset version

scale up to 120w

this version is ready to scale up without too much ridiculus bugs

v0.5

The version ready to scale up to arxix papers

Support Reading Annotation

add reading annotations

acceptable annotation result

This version can correctly generate the bounding box for table, equation and algorithm envs.