Emergent modality binding (vi -> te -> au) on MSRVTT (test)
17.4mAPVCLIP+CLAP+BB
Evaluation results
| Method | Links | |
|---|---|---|
| VCLIP+CLAP+BBSource Dataset=AVE, Evaluation Protocol=vi -> te -> au2026.02 | 17.4 | |
| Ex-MCRSource Dataset=AVE, Evaluation Protocol=vi -> te -> au2026.02 | 10.2 | |
| C-MCRSource Dataset=AVE, Evaluation Protocol=vi -> te -> au2026.02 | 9.8 | |
| VCLIP+CLAP+FTSource Dataset=AVE, Evaluation Protocol=vi -> te -> au2026.02 | 9.6 | |
| VCLIP+CLAPSource Dataset=AVE, Evaluation Protocol=vi -> te -> au2026.02 | 6.2 |