FEAT: Full-Dimensional Efficient Attention Transformer for Medical Video Generation Paper • 2506.04956 • Published Jun 5 • 3