| Benchmark | Model | Rank | Results |
|---|---|---|---|
| zero-shot-question-answering-on-sciq | GPT-NeoX-20B (LM Evaluation Harness) | #2 | Accuracy: 92.8 |
| common-sense-reasoning-on-arc-easy | GPT-NeoX-20B (5-shot) | #37 | Accuracy: 74.6 |
| multi-task-language-understanding-on-mmlu | GPT-NeoX 20B (5-shot) | #81 | Average (%): 33.6 |
| common-sense-reasoning-on-winogrande | GPT-NeoX-20B (5-shot) | #83 | Accuracy: 68.3 |
| question-answering-on-piqa | GPT-NeoX-20B (5-shot) | #86 | Accuracy: 77.4 |
| sentence-completion-on-hellaswag | GPT-NeoX-20B (5-shot) | #168 | Accuracy: 53.8 |