Natural Language Understanding on GLUE (train)
95SST-2 AccuracyFine-tuning
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Fine-tuningBackbone=RoBERTa-large, Training size=Full training set2021.08 | 95 | 89.8 | 93.3 | 91.4 | 81.7 | |
| LM-BFFBackbone=RoBERTa-large, Training size=Full training set, Demonstration usage=without demonstration2021.08 | 94.9 | 89.6 | 92.8 | 91.7 | 86.4 | |
| DARTBackbone=RoBERTa-large, Training size=Full training set2021.08 | 94.6 | 87.3 | 92.3 | 90.4 | 89.5 |