Multimodal Classification on MM-IMDB-Hard (test)
68.2Macro F1MMBT-Large
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MMBT-LargeModel Scale=Large2019.09 | 68.2 | 70.3 | |
| ViLBertFine-tuning Source=Standard pre-trained2019.09 | 65.4 | 68.6 | |
| MMBTModel Scale=Base2019.09 | 65.3 | 68.6 | |
| ViLBert-VCRFine-tuning Source=VCR2019.09 | 63.4 | 66.9 | |
| ViLBert-RefcocoFine-tuning Source=Refcoco2019.09 | 63.4 | 67.1 | |
| ViLBert-Flickr30kFine-tuning Source=Flickr30k2019.09 | 63.4 | 67 | |
| ViLBert-VQAFine-tuning Source=VQA2019.09 | 62.7 | 66.2 |