Natural Language Processing on SuperGLUE Full, excl. ReCoRD (dev)
70.03Macro Avg ScoreMulti-CLS BERT
Evaluation Results
| Method | Links | |
|---|---|---|
| Multi-CLS BERTBackbone=BERT Large, Model Size=350.9M, K=5, lambda=0.12022.10 | 70.03 | |
| Multi-CLS BERTBackbone=BERT Large, Model Size=350.9M, K=5, lambda=0.52022.10 | 69.79 | |
| Multi-CLS BERTBackbone=BERT Large, Model Size=350.9M, K=5, lambda=02022.10 | 69.56 | |
| Multi-CLS BERTBackbone=BERT Large, Model Size=338.3M, K=12022.10 | 69.24 | |
| MTLBackbone=BERT Large, Model Size=335.2M2022.10 | 69.16 | |
| Multi-CLS BERTBackbone=BERT Large, Model Size=350.9M, K=5, lambda=12022.10 | 69.04 | |
| Multi-CLS BERTBackbone=BERT Base, Model Size=118.4M, K=5, lambda=02022.10 | 66.8 | |
| Multi-CLS BERTBackbone=BERT Base, Model Size=118.4M, K=5, lambda=0.52022.10 | 66.8 | |
| Multi-CLS BERTBackbone=BERT Base, Model Size=118.4M, K=5, lambda=0.12022.10 | 66.74 | |
| Multi-CLS BERTBackbone=BERT Base, Model Size=118.4M, K=5, lambda=12022.10 | 66.39 | |
| MTLBackbone=BERT Base, Model Size=109.5M2022.10 | 66.33 | |
| Multi-CLS BERTBackbone=BERT Base, Model Size=111.3M, K=12022.10 | 66.29 | |
| PretrainedBackbone=BERT Base, Model Size=109.5M2022.10 | 65.04 |