Reasoning on BigBench Hard Penguins
44.1AccuracyReElicit
Evaluation results
| Method | Links | |
|---|---|---|
| ReElicitevaluations=30 prompt evaluations2026.05 | 44.1 | |
| OPROevaluations=30 prompt evaluations2026.05 | 43.9 | |
| APEevaluations=30 prompt evaluations2026.05 | 43.4 | |
| TextGradevaluations=30 prompt evaluations2026.05 | 33.1 | |
| PromptBreederevaluations=30 prompt evaluations2026.05 | 29.3 |