| Benchmark | Model | Rank | Results |
|---|---|---|---|
| question-answering-on-boolq | Self-Improving Pretraining (1.4B) | – | Accuracy: 70.3 |
| question-answering-on-piqa | Self-Improving Pretraining (1.4B) | – | Accuracy: 75.8 |
| sentence-completion-on-hellaswag | Self-Improving Pretraining (1.4B) | – | Accuracy: 69.4 |