Datasets:
Add aligned English and Chinese UMBench data with images
#1
by gaviZh - opened
Add 581 shared PNG images and 2,728 image-backed questions per language. Preserve all 2,761 original annotations per language, document 33 questions with missing images, flag 5 invalid answer labels, and provide bilingual loading instructions.
gaviZh changed pull request status to open
libo200404 changed pull request status to merged