Natural Language Understanding on GLUE (val test)
0.97SST-2 AccuracyXLNet
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| XLNetPre-training Strategy=Natural Language2023.02 | 0.97 | 0.859 | 0.908 | 0.923 | 0.908 | 0.949 | |
| ELECTRAPre-training Strategy=Natural Language2023.02 | 0.969 | 0.88 | 0.908 | 0.924 | 0.909 | 0.95 | |
| UNIMOPre-training Strategy=Multimodal2023.02 | 0.968 | — | — | — | 0.898 | 0.948 | |
| DeBERTaPre-training Strategy=Natural Language2023.02 | 0.968 | 0.883 | 0.919 | 0.923 | 0.911 | 0.953 | |
| OFAPre-training Strategy=Multimodal2023.02 | 0.966 | 0.91 | 0.917 | 0.925 | 0.902 | 0.945 | |
| RoBERTaPre-training Strategy=Natural Language2023.02 | 0.964 | 0.866 | 0.909 | 0.922 | 0.902 | 0.939 | |
| mPLUG-2Deberta2023.02 | 0.962 | 0.894 | 0.921 | 0.926 | 0.908 | 0.948 | |
| mPLUG-22023.02 | 0.951 | 0.88 | 0.901 | 0.927 | 0.902 | 0.945 | |
| mPLUG-2Base2023.02 | 0.935 | 0.852 | 0.873 | 0.913 | 0.876 | 0.932 | |
| BERTPre-training Strategy=Natural Language2023.02 | 0.932 | 0.704 | 0.88 | 0.913 | 0.866 | 0.923 | |
| SimVLMPre-training Strategy=Multimodal2023.02 | 0.909 | 0.639 | 0.752 | 0.904 | 0.834 | 0.886 | |
| FLAVAPre-training Strategy=Multimodal2023.02 | 0.909 | 0.578 | 0.814 | 0.904 | 0.803 | 0.873 | |
| ViLBERTPre-training Strategy=Multimodal2023.02 | 0.904 | 0.537 | 0.69 | 0.886 | 0.799 | 0.838 | |
| LXMERTPre-training Strategy=Multimodal2023.02 | 0.902 | 0.572 | 0.698 | 0.753 | 0.804 | 0.842 | |
| Uni-PerceiverPre-training Strategy=Multimodal2023.02 | 0.902 | 0.643 | 0.866 | 0.871 | 0.817 | 0.899 | |
| VL-BERTPre-training Strategy=Multimodal2023.02 | 0.898 | 0.557 | 0.706 | 0.89 | 0.812 | 0.863 | |
| UNITERPre-training Strategy=Multimodal2023.02 | 0.897 | 0.556 | 0.693 | 0.892 | 0.809 | 0.86 | |
| VisualBERTPre-training Strategy=Multimodal2023.02 | 0.894 | 0.566 | 0.719 | 0.894 | 0.816 | 0.87 |