Mathematical Reasoning on AIME 25 (Pass@32)
43.3Pass@32OPEFO
Evaluation Results
| Method | Links | |
|---|---|---|
| OPEFOBackbone=Qwen2.5-Math-7B2026.05 | 43.3 | |
| Clip-CovBackbone=Qwen2.5-Math-7B2026.05 | 41.7 | |
| GRPO (Strict)Backbone=Qwen2.5-Math-7B2026.05 | 41.2 | |
| KL-CovBackbone=Qwen2.5-Math-7B2026.05 | 40.9 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.05 | 38.4 | |
| Clip-higherBackbone=Qwen2.5-Math-7B2026.05 | 37.6 | |
| Entropy-RegBackbone=Qwen2.5-Math-7B2026.05 | 33.6 |