Fetching the paper…

VATT: Transformers for Multimodal Self-Supervised Learning from Raw Video, Audio and Text · Around