Search papers, labs, and topics across Lattice.
1
0
3
Superpixel Transformers show that strategically chunking images into superpixels can boost performance over standard Vision Transformers, suggesting a new path for efficient and effective attention mechanisms.
