| Benchmark | Model | Rank | Results |
|---|---|---|---|
| common-sense-reasoning-on-arc-easy | GPT-SW3 40B | – | Accuracy: 68.7 |
| common-sense-reasoning-on-winogrande | GPT-SW3 40B | – | Accuracy: 65.6 |
| question-answering-on-piqa | GPT-SW3 40B | – | Accuracy: 77.2 |
| sentence-completion-on-hellaswag | GPT-SW3 40B | – | Accuracy: 53.2 |
| zero-shot-question-answering-on-openbookqa | GPT-SW3 40B | – | Accuracy: 27.4 |