Search papers, labs, and topics across Lattice.
Affiliation:
1
0
3
DSAQuant reveals that aligning quantization training with the stage-wise nature of video diffusion can drastically enhance visual fidelity in text-to-video generation.
