Mathematical Reasoning on GSM8K (Out-of-distribution)
92AccuracyRLTR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RLTRSamples for Acc.=642026.02 | 92 | 93.4 | 94 | 94.2 | |
| RLVRSamples for Acc.=642026.02 | 89.4 | 90.9 | 92.5 | 92.9 | |
| Base modelSamples for Acc.=642026.02 | 89.1 | 91.5 | 92.7 | 93.3 |