Search for a command to run...
Video-LaVIT: Unified Video-Language Pre-training with Decoupled Visual-Motional Tokenization