Mathematical Reasoning on GSM8K (test) (Avg@8)
91.5Accuracy (Avg@8)SAERLG
Evaluation Results
| Method | Links | |
|---|---|---|
| SAERLGModel=Qwen2.5-Math-7B, Evaluation Step=9002026.05 | 91.5 | |
| DIFFModel=Qwen2.5-Math-7B, Evaluation Step=9002026.05 | 90.9 | |
| GRPOModel=Qwen2.5-Math-7B, Evaluation Step=9002026.05 | 90.3 | |
| GAINRLModel=Qwen2.5-Math-7B, Evaluation Step=9002026.05 | 90.1 | |
| ADARFTModel=Qwen2.5-Math-7B, Evaluation Step=9002026.05 | 87.5 | |
| SAERLDModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 84.6 | |
| SAERLGModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 83.5 | |
| DAPOModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 81.9 | |
| DIFFModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 81.6 | |
| GRPOModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 81.1 | |
| GAINRLModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 79.2 | |
| ADARFTModel=Qwen2.5-Math-1.5B, Evaluation Step=9002026.05 | 78.9 |