Mathematical Reasoning on Olympiad (Mean@32 accuracy)
29.86Mean@32 AccuracyGRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| GRPOModel=Qwen2.5-Math-7B, G=22026.06 | 29.86 | |
| GRPOModel=Qwen2.5-Math-7B, G=162026.06 | 28.66 | |
| C-RF w/ NTFModel=Qwen2.5-Math-7B, G=12026.06 | 26.54 | |
| RF++ w/ NTFModel=Qwen2.5-Math-7B, G=12026.06 | 24.92 | |
| GRPOModel=Qwen2.5-Math-1.5B, G=162026.06 | 24.86 | |
| RF++Model=Qwen2.5-Math-7B, G=82026.06 | 24.74 | |
| C-RF w/ NTFModel=Qwen2.5-Math-1.5B, G=12026.06 | 23.89 | |
| GRPOModel=Qwen2.5-Math-1.5B, G=22026.06 | 23.52 | |
| RF++Model=Qwen2.5-Math-1.5B, G=82026.06 | 22.87 | |
| RF++ w/ NTFModel=Qwen2.5-Math-1.5B, G=12026.06 | 21.68 |