Concept Detection on AG News (test)
75.9AccuracyEnsembling (Boosting)
Evaluation results
| Method | Links | |
|---|---|---|
| Ensembling (Boosting)Base Model=Pythia-70M, Ensemble Size=8, Number of top concept-associated features=12025.05 | 75.9 | |
| Expanded SAEBase Model=Pythia-70M, Ensemble Size=1, Number of top concept-associated features=12025.05 | 74.6 | |
| Ensembling (NB)Base Model=Pythia-70M, Ensemble Size=8, Number of top concept-associated features=12025.05 | 74.2 |