agianbig/hardblink_eval download history

agianbig/hardblink_eval is a visual question answering dataset on the Hugging Face Hub. In the last 30 days it was downloaded 39 times (9 in the last 7 days), and 549 times in total. It ranks #223,370 among datasets by monthly downloads.

HardBLINK: Relative-Depth Evaluation Evaluation set for Ablate-to-Validate: Are Vision-Language Models Really Using Visual Reasoning Tokens? (code · project page · arXiv). A harder relative-depth probe in the BLINK style: each image has 3, 4, or 5 labeled points circled, and the task

Open agianbig/hardblink_eval on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.