Search papers, labs, and topics across Lattice.
1
0
3
Forget slow, white-box optimization: VQQA uses a multi-agent system to ask questions about generated videos, using the answers as semantic gradients to drastically improve video quality via black-box prompt refinement.
