Mathematical Reasoning on MAWPS Out-of-Distribution (test)
96.5AccuracyCausalRM
Evaluation Results
| Method | Links | |
|---|---|---|
| CausalRMEvaluation Type=Downstream RLHF2026.01 | 96.5 | |
| GoalRMEvaluation Type=Downstream RLHF2026.01 | 95.9 | |
| Standard RMEvaluation Type=Downstream RLHF2026.01 | 95.3 | |
| SFTEvaluation Type=Downstream RLHF2026.01 | 93.6 | |
| InfoRMEvaluation Type=Downstream RLHF2026.01 | 51.4 |