README.md

metadata

task_categories:
  - visual-question-answering
pretty_name: chartqa_pro_lite

gsarch/chartqa_pro_lite

A deterministic lite evaluation subset of ahmed-masry/ChartQAPro.

• Source revision: e27c2874825874d6767d2bbc538ed4f0dc2c64c2
• Source split: test
• Sampling seed: 45
• Output rows: 500
• Schema: unchanged from the upstream dataset

Generated by scripts/create_lite_eval_datasets.py in the gaze-vlm vero_eval repository.