Mathematical Reasoning on Minerva (pass@3)
56.68pass@3HDPO
Evaluation results
| Method | Links | |
|---|---|---|
| HDPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 56.68 | |
| HDPOBackbone=Qwen3-4B, Number of samples=162026.06 | 54.92 | |
| HDPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 53.54 | |
| Critique-GRPOBackbone=8B, Number of samples=162026.06 | 51.51 | |
| EDGE-GRPOBackbone=Qwen-7B, Number of samples=162026.06 | 49.89 | |
| GRPOBackbone=Qwen3-4B, Number of samples=162026.06 | 49.17 | |
| GRPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 45.14 | |
| GRPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 44.86 | |
| LUFFYBackbone=Qwen-Math-7B, Number of samples=162026.06 | 44.54 | |
| SimpleRLBackbone=Qwen-2.5-7B, Number of samples=162026.06 | 40.43 | |
| Base ModelBackbone=Qwen3-4B, Number of samples=162026.06 | 39.82 | |
| OatBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 38.75 | |
| Base ModelBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 37.67 | |
| Base ModelBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 30.17 |