Mathematical Reasoning on OlympiadBench (pass@3)
64.76Pass@3HDPO
Evaluation Results
| Method | Links | |
|---|---|---|
| HDPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 64.76 | |
| HDPOBackbone=Qwen3-4B, Number of samples=162026.06 | 62.37 | |
| Critique-GRPOBackbone=8B, Number of samples=162026.06 | 61.46 | |
| HDPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 56.19 | |
| LUFFYBackbone=Qwen-Math-7B, Number of samples=162026.06 | 56.09 | |
| GRPOBackbone=Qwen3-4B, Number of samples=162026.06 | 55.5 | |
| GRPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 54.74 | |
| GRPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 53.68 | |
| OatBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 50.82 | |
| SimpleRLBackbone=Qwen-2.5-7B, Number of samples=162026.06 | 49.84 | |
| EDGE-GRPOBackbone=Qwen-7B, Number of samples=162026.06 | 48.79 | |
| Base ModelBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 43.73 | |
| Base ModelBackbone=Qwen3-4B, Number of samples=162026.06 | 40.11 | |
| Base ModelBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 38.88 |