Computation Cost Analysis on Reasoning Samples 10K
0.16Time RatioCPMI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CPMIReward Type=CPMI2026.04 | 0.16 | 0.02 | |
| CPMI_MergeReward Type=CPMI_Merge2026.04 | 0.3 | 0.18 | |
| MCReward Type=MC2026.04 | 1 | 1 | |
| PAVReward Type=PAV2026.04 | 1.17 | 2.38 |