Hugging Face daily papers·6d agoTT-VidT: Decoupling the Temporal Axis for Efficient Motion-Centric Video Pretraining#video-pretraining#self-supervised-learning#dinov3