McGill-NLP/SpeechJBB download history
McGill-NLP/SpeechJBB is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 1,618 times (67 in the last 7 days), and 29,548 times in total. It ranks #19,755 among datasets by monthly downloads.
SpeechJBB Dataset Description SpeechJBB is an audio benchmark for evaluating safety alignment and comprehension in large audio language models (LALMs) under multilingual and code-switched speech. It is designed to test whether models consistently refuse harmful spoken r
Spaces using SpeechJBB
- Model Pulse 8 likes
Open McGill-NLP/SpeechJBB on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.