Search papers, labs, and topics across Lattice.
1
0
3
11
Image editing benchmarks are broken: even GPT-4 is worse than the new PVC-Judge model at assessing visual consistency in edited images.
