Search papers, labs, and topics across Lattice.
1
0
3
5
Achieve 25x faster audio-visual generation without sacrificing quality by distilling a bidirectional diffusion model into a streaming autoregressive model, unlocking real-time applications.
