5 ms·Combining next-token prediction and video diffusion in computer vision, robotics2 points by cubefox 2y ago