Search papers, labs, and topics across Lattice.
Zhejiang University
1
0
3
Interactive video world models can be sped up by 2.5x without retraining, simply by being smarter about how they use context and computation based on the user's actions.
