Search papers, labs, and topics across Lattice.
1
0
3
22
Medical vision-language models perform better when the modality gap is tuned to an intermediate level, challenging the assumption that minimizing it is always optimal.
