Math Reasoning on MATH 500 (Score %)
90.2MATH 500 ScoreSoft-RLVR
Evaluation Results
| Method | Links | |
|---|---|---|
| Soft-RLVRVerifier=GPT-OSS-120B, Checklist source=DeepSeek R1, Training step=5002026.05 | 90.2 | |
| Soft-RLVRVerifier=GPT-OSS-20B, Checklist source=DeepSeek R1, Training step=5002026.05 | 89.6 | |
| Oracle rule-basedReward=Oracle rule-based, Training step=5002026.05 | 87.8 | |
| Command R7B (No RL)Reward=None (Initial Policy), Training step=5002026.05 | 87.6 |