Visual Question Answering on CLEVR-CoGenT (in-distribution)
98AccuracyVector-NMN
Evaluation results
| Method | Links | |
|---|---|---|
| Vector-NMNNMN architecture=Vector-NMN, Image encoder type=all2021.06 | 98 | |
| Tensor-NMNNMN architecture=Tensor-NMN, Image encoder type=all2021.06 | 97.9 | |
| Vector-NMNNMN architecture=Vector-NMN, Image encoder type=group2021.06 | 94.4 |