Mathematical Reasoning on MATH 500 (P@1)
83.8P@1GRPO w/ OAR-P
Evaluation Results
| Method | Links | |
|---|---|---|
| GRPO w/ OAR-PBackbone=Qwen2.5-Math-7B2026.01 | 83.8 | |
| GRPO w/ OAR-GBackbone=Qwen2.5-Math-7B2026.01 | 83 | |
| GRPO w/ KTAEBackbone=Qwen2.5-Math-7B2026.01 | 82.4 | |
| GRPO w/ Entropy AdvBackbone=Qwen2.5-Math-7B2026.01 | 81.7 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.01 | 80.8 | |
| GRPO w/ Random AdvBackbone=Qwen2.5-Math-7B2026.01 | 79.8 | |
| GRPO w/ OAR-GBackbone=Qwen2.5-7B2026.01 | 78.7 | |
| GRPO w/ OAR-PBackbone=Qwen2.5-7B2026.01 | 78.4 | |
| GRPO w/ Entropy AdvBackbone=Qwen2.5-7B2026.01 | 77.2 | |
| GRPOBackbone=Qwen2.5-7B2026.01 | 75.8 | |
| GRPO w/ Random AdvBackbone=Qwen2.5-7B2026.01 | 75.2 | |
| Qwen2.5-7BBackbone=Qwen2.5-7B2026.01 | 52.7 | |
| Qwen2.5-Math-7BBackbone=Qwen2.5-Math-7B2026.01 | 51.2 |