Mathematical Reasoning on Aggregate AIME, AMC, GSM8K, MATH, MNV, OLPD
173Avg Training Steps to Target AccSAERLG
Evaluation Results
| Method | Links | |
|---|---|---|
| SAERLGModel=Qwen2.5-Math-7B2026.05 | 173 | |
| DIFFModel=Qwen2.5-Math-7B2026.05 | 180 | |
| GAINRLModel=Qwen2.5-Math-7B2026.05 | 193 | |
| GRPOModel=Qwen2.5-Math-7B2026.05 | 200 | |
| SAERLDModel=Qwen2.5-Math-1.5B2026.05 | 206 | |
| DAPOModel=Qwen2.5-Math-1.5B2026.05 | 330 | |
| DIFFModel=Qwen2.5-Math-1.5B2026.05 | 373 | |
| SAERLGModel=Qwen2.5-Math-1.5B2026.05 | 380 | |
| GRPOModel=Qwen2.5-Math-1.5B2026.05 | 470 | |
| ADARFTModel=Qwen2.5-Math-7B2026.05 | 476 | |
| GAINRLModel=Qwen2.5-Math-1.5B2026.05 | 523 | |
| ADARFTModel=Qwen2.5-Math-1.5B2026.05 | 676 |