Search papers, labs, and topics across Lattice.
The Hong Kong Polytechnic University, OPPO Research Institute
2
0
4
Decoupling memory conditioning from video generation allows for more data-efficient training and better spatial consistency in long-horizon video generation, even when exploring novel scenes.
BinaryAttention proves you can more than halve the runtime of attention in vision and diffusion transformers without sacrificing accuracy, simply by using the sign of queries and keys.
