Mathematical Reasoning on Math (Accuracy, Inference Latency)
93.28AccuracyHDPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HDPOBackbone=Qwen2.5-Math-7B2026.06 | 93.28 | 0.16 | |
| Critique-GRPOBackbone=Qwen2.5-Math-7B2026.06 | 91.22 | 0.19 | |
| GRPOBackbone=Qwen2.5-Math-7B2026.06 | 89.72 | 0.12 | |
| LUFFYBackbone=Qwen2.5-Math-7B2026.06 | 87.99 | 0.14 | |
| OatBackbone=Qwen2.5-Math-7B2026.06 | 85.34 | 0.07 | |
| SimpleRLBackbone=Qwen2.5-Math-7B2026.06 | 85.22 | 0.09 | |
| EDGE-GRPOBackbone=Qwen2.5-Math-7B2026.06 | 84.1 | 0.07 |