Mathematical Computation on DAPO Math
26.56Exact Match (EM)Prompt-R1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Prompt-R1Backbone=GPT-4o-mini2025.11 | 26.56 | 26.56 | |
| CoT ReasoningBackbone=GPT-4o-mini2025.11 | 20.31 | 20.32 | |
| BaselineBackbone=GPT-4o-mini2025.11 | 18.75 | 18.76 | |
| GEPAOptimization Framework=APO, Backbone=GPT-4o-mini2025.11 | 13.28 | 14.06 | |
| TextGradOptimization Framework=APO, Backbone=GPT-4o-mini2025.11 | 10.16 | 10.27 | |
| OPROOptimization Framework=APO, Backbone=GPT-4o-mini2025.11 | 6.25 | 6.39 | |
| GRPOBackbone=Qwen3-4B2025.11 | 3.91 | 3.91 | |
| SFTBackbone=Qwen3-4B2025.11 | 3.13 | 3.13 | |
| BaselineBackbone=Qwen3-4B2025.11 | 0 | 0 | |
| CoT ReasoningBackbone=Qwen3-4B2025.11 | 0 | 0 |