Mathematical Reasoning on MATH-500 (train)
45.2AccuracyJustGRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| JustGRPOSeq Len=5122026.01 | 45.2 | |
| JustGRPOSeq Len=2562026.01 | 45.1 | |
| ESPOSeq Len=5122026.01 | 43.4 | |
| SPGSeq Len=5122026.01 | 41.8 | |
| GDPOSeq Len=5122026.01 | 41.4 | |
| LLaDOUSeq Len=2562026.01 | 41.1 | |
| d1Seq Len=5122026.01 | 40.2 | |
| SPGSeq Len=2562026.01 | 40 | |
| GDPOSeq Len=2562026.01 | 39.6 | |
| wd1Seq Len=5122026.01 | 39 | |
| ESPOSeq Len=2562026.01 | 39 | |
| JustGRPOSeq Len=1282026.01 | 39 | |
| d-TreeRPOSeq Len=5122026.01 | 38.9 | |
| d1Seq Len=2562026.01 | 38.6 | |
| d-TreeRPOSeq Len=2562026.01 | 37.7 | |
| ESPOSeq Len=1282026.01 | 36 | |
| wd1Seq Len=2562026.01 | 34.4 | |
| d1Seq Len=1282026.01 | 33.8 | |
| SPGSeq Len=1282026.01 | 33.4 | |
| wd1Seq Len=1282026.01 | 33.3 | |
| GDPOSeq Len=1282026.01 | 33.2 |