Reasoning on PUPA (test)
91.85ScoreGEPA
Evaluation Results
| Method | Links | |
|---|---|---|
| GEPABackbone=Qwen3 8B, Optimization Budget (# rollouts)=24262025.07 | 91.85 | |
| GRPOBackbone=Qwen3 8B, Optimization Budget (# rollouts)=240002025.07 | 86.66 | |
| GEPA+MergeBackbone=Qwen3 8B, Optimization Budget (# rollouts)=24262025.07 | 86.26 | |
| MIPROv2Backbone=Qwen3 8B2025.07 | 81.55 | |
| BaselineBackbone=Qwen3 8B2025.07 | 80.82 |