Mathematical Reasoning on PRMBench
60.7PRMScoreCPMI_Merge
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPMI_MergeReward Type=CPMI_Merge2026.04 | 60.7 | 5.24 | |
| CPMIReward Type=CPMI2026.04 | 58.8 | 9.53 | |
| PAVReward Type=PAV2026.04 | 49.6 | 1.09 | |
| MCReward Type=MC2026.04 | 38.8 | 1 |
Loading the SOTA2 catalog…