General Reasoning on GPQA Diamond (pass@3)
69.89pass@3HDPO
Evaluation results
| Method | Links | |
|---|---|---|
| HDPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 69.89 | |
| Critique-GRPOBackbone=8B, Number of samples=162026.06 | 67.28 | |
| HDPOBackbone=Qwen3-4B, Number of samples=162026.06 | 65.49 | |
| HDPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 63.41 | |
| LUFFYBackbone=Qwen-Math-7B, Number of samples=162026.06 | 61.51 | |
| GRPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 60.18 | |
| GRPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 58.6 | |
| GRPOBackbone=Qwen3-4B, Number of samples=162026.06 | 58.47 | |
| Base ModelBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 48.72 | |
| SimpleRLBackbone=Qwen-2.5-7B, Number of samples=162026.06 | 47.31 | |
| Base ModelBackbone=Qwen3-4B, Number of samples=162026.06 | 46.81 | |
| EDGE-GRPOBackbone=Qwen-7B, Number of samples=162026.06 | 46.76 | |
| OatBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 44.43 | |
| Base ModelBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 41.29 |