| Benchmark | Model | Rank | Results |
|---|---|---|---|
| common-sense-reasoning-on-winogrande | BaseLoop (Llama3.1-1B, 3x4, Muon) | – | Accuracy: 52.88 |
| question-answering-on-piqa | BaseLoop (Llama3.1-1B, 3x4, Muon) | – | Accuracy: 65.18 |
| reading-comprehension-on-race | BaseLoop (Llama3.1-1B, 3x4, Muon) | – | Accuracy: 30.62 |
| sentence-completion-on-hellaswag | BaseLoop (Llama3.1-1B, 3x4, Muon) | – | Accuracy: 31.48 |
| zero-shot-question-answering-on-openbookqa | BaseLoop (Llama3.1-1B, 3x4, Muon) | – | Accuracy: 20.8 |