Search papers, labs, and topics across Lattice.
University of California
1
0
4
Forget training costly reward models for text-to-image alignment – AutoRubric-T2I learns interpretable rubrics that outperform them using less than 0.01% of the data.
