Sentiment Classification on SST-2 (OOD)
82.4AccuracyDynamic
Evaluation results
| Method | Links | |
|---|---|---|
| DynamicTraining set size=800 total, Backbone=RoBERTa2022.03 | 82.4 | |
| Dynamic-MRTraining set size=800 total, Backbone=RoBERTa2022.03 | 82.24 | |
| Dynamic-FRTraining set size=800 total, Backbone=RoBERTa2022.03 | 81.59 | |
| Static + 350 autoTraining set size=400, Backbone=RoBERTa-base2022.03 | 81.26 | |
| Static + 350 autoTraining set size=400 total, Backbone=RoBERTa2022.03 | 81.26 | |
| Static + 550 autoTraining set size=600, Backbone=RoBERTa-base2022.03 | 81.23 | |
| Static + 750 autoTraining set size=800, Backbone=RoBERTa-base2022.03 | 80.75 | |
| Automatics CADTraining set size=3,414 total, Backbone=RoBERTa2022.03 | 80.6 | |
| Static + 950 autoTraining set size=1000, Backbone=RoBERTa-base2022.03 | 80.58 | |
| ALTraining set size=100 gold, Backbone=RoBERTa2022.03 | 80.5 | |
| Manual CADTraining set size=3,414 gold, Backbone=RoBERTa2022.03 | 80.3 | |
| FullTraining set size=1707, Backbone=RoBERTa-base2022.03 | 80.23 | |
| Static + 1150 autoTraining set size=1200, Backbone=RoBERTa-base2022.03 | 79.52 | |
| Static + 150 autoTraining set size=200, Backbone=RoBERTa-base2022.03 | 79.4 | |
| StaticTraining set size=50, Backbone=RoBERTa-base2022.03 | 79.29 | |
| StaticTraining set size=50 gold, Backbone=RoBERTa2022.03 | 79.29 | |
| RRTraining set size=400, Backbone=RoBERTa-base2022.03 | 78.89 | |
| DPTraining set size=400, Backbone=RoBERTa-base2022.03 | 77.33 |