Math Reasoning on Open-RS
98.47Relative Overall ScoreJensen-Shannon
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Jensen-ShannonBackbone=Qwen2.5-Math-7B2026.02 | 98.47 | 2 | |
| Reverse KLBackbone=Qwen2.5-Math-7B2026.02 | 96.78 | 2.4 | |
| KLBackbone=Qwen2.5-Math-7B2026.02 | 95.16 | 3 | |
| Jensen-ShannonBackbone=Qwen2.5-Math-1.5B2026.02 | 93.3 | 2.4 | |
| PearsonBackbone=Qwen2.5-Math-7B2026.02 | 91.86 | 5 | |
| HellingerBackbone=Qwen2.5-Math-7B2026.02 | 91.7 | 4.4 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.02 | 90.52 | 5.8 | |
| Total VariationBackbone=Qwen2.5-Math-1.5B2026.02 | 86.93 | 4 | |
| Total VariationBackbone=Qwen2.5-Math-7B2026.02 | 86.05 | 4.8 | |
| KLBackbone=Qwen2.5-Math-1.5B2026.02 | 85.66 | 4 | |
| PearsonBackbone=Qwen2.5-Math-1.5B2026.02 | 85.19 | 3.6 | |
| GRPOBackbone=Qwen2.5-Math-1.5B2026.02 | 82.44 | 4.6 | |
| Reverse KLBackbone=Qwen2.5-Math-1.5B2026.02 | 81.65 | 5.2 | |
| HellingerBackbone=Qwen2.5-Math-1.5B2026.02 | 78.47 | 3.6 | |
| BaseBackbone=Qwen2.5-Math-1.5B2026.02 | 0 | 8 | |
| BaseBackbone=Qwen2.5-Math-7B2026.02 | 0 | 8 |