Sentiment Classification on BiasedSST (test)
95.9Average AccuracyQ-Diversity
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Q-Diversity2023.05 | 95.9 | 68.2 | |
| ERM2023.05 | 95.1 | 2.15 | |
| CVaR DRO2023.05 | 92.5 | 28.1 | |
| JTTSupervision=unsupervised2023.05 | 84.2 | 35 | |
| Oracle DROSupervision=fully-supervised2023.05 | 77.9 | 67.7 |