Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
CapProbe reveals that many vision-language models exhibit substantial gaps in coverage, challenging the reliability of existing caption evaluation metrics.
