Mathematical Reasoning on Average (Math-500, GSM8K, Big-Math)
72.9AccuracyRLVR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RLVRTraining Set=Big-Math2025.07 | 72.9 | 0.47 | 0.28 | 0.26 | |
| RLVR + BCE ClassifierTraining Set=Big-Math2025.07 | 72.9 | 0.78 | 0.15 | 0.1 | |
| RLVR + Brier ClassifierTraining Set=Big-Math2025.07 | 72.9 | 0.78 | 0.15 | 0.1 | |
| RLVR + ProbeTraining Set=Big-Math2025.07 | 72.9 | 0.65 | 0.19 | 0.13 | |
| Answer ProbabilityTraining Set=Big-Math2025.07 | 72.9 | 0.52 | 0.26 | 0.26 | |
| RLCRTraining Set=Big-Math2025.07 | 72.7 | 0.67 | 0.17 | 0.1 | |
| SFT+RLCRTraining Set=Big-Math2025.07 | 72.2 | 0.78 | 0.14 | 0.08 | |
| BaseTraining Set=Big-Math2025.07 | 56.1 | 0.56 | 0.4 | 0.39 |