Mathematical Reasoning on LiveBench Math (test)
51.95ScoreGEPA
Evaluation Results
| Method | Links | |
|---|---|---|
| GEPABackbone=Qwen3 8B, Optimization Budget (# rollouts)=18392025.07 | 51.95 | |
| GEPA+MergeBackbone=Qwen3 8B, Optimization Budget (# rollouts)=18392025.07 | 51.95 | |
| GRPOBackbone=Qwen3 8B, Optimization Budget (# rollouts)=240002025.07 | 51.26 | |
| BaselineBackbone=Qwen3 8B2025.07 | 48.7 | |
| MIPROv2Backbone=Qwen3 8B2025.07 | 46.6 |