gwkrsrch/avqa_hard download history
gwkrsrch/avqa_hard is a video text to text dataset on the Hugging Face Hub. In the last 30 days it was downloaded 67 times (11 in the last 7 days), and 116 times in total. It ranks #206,837 among datasets by monthly downloads.
AVQA-Hard AVQA-Hard: the 1,696 items of gwkrsrch/avqa_2025 that GPT-4o answers incorrectly when shown only the central frame of the clip, without audio. Released with Do Modern Video-LLMs Need to Listen? A Benchmark Audit and Scalable Remedy (Interspeech 2026, arXiv:2509.17901); code:
Open gwkrsrch/avqa_hard on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.