Prompt Optimization on 10-task prompt optimization suite GSM8K MMLU BBH
81Average Win/Tie RateReElicit
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ReElicitNumber of prompt evaluations=302026.05 | 81 | — | 78 | 79 | 85 | 82 | |
| OPRONumber of prompt evaluations=302026.05 | 58 | 31 | 66 | — | 71 | 63 | |
| APENumber of prompt evaluations=302026.05 | 56 | 30 | — | 60 | 71 | 63 | |
| TextGradNumber of prompt evaluations=302026.05 | 49 | 26 | 48 | 51 | 70 | — | |
| PromptBreederNumber of prompt evaluations=302026.05 | 40 | 22 | 44 | 45 | — | 48 |