kaze-desu/

imagenet-r-test

This dataset has been converted to VQA format with:

problem: Question with multiple-choice options (format: "What type of object is in the photo?\n Please choose one from list [ ... ].") solution: Answer in classname format split: 'test'

Original dataset: imagenet-r-test Total samples: 30000 Number of classes: 200 Choices per question: 80 Format: Parquet (consistent with original dataset)

from… See the full description on the dataset page: https://huggingface.co/datasets/kaze-desu/imagenet-r-test.