ai.meta.com
Introducing MUGEN, a new dataset for multimodal research
To help advance research on AI that works across video, audio, text and more, we are releasing the MUGEN dataset. It contains 375K samples w/rich annotations specifically designed to train AI on...
