Search papers, labs, and topics across Lattice.
3
0
7
6
FlashDecoder decodes video frames 3.6x-4.7x faster than traditional methods while maintaining high-quality reconstruction at 1080p.
EventCoT achieves state-of-the-art RTL performance with fewer visual tokens, revolutionizing how we approach temporal reasoning in videos.
Cycle action consistency in planning can drastically reduce computational costs while maintaining accuracy across diverse manipulation and navigation tasks.
