Mathematical Reasoning on AMC 2023 (pass@3)
85.79pass@3HDPO
Evaluation Results
| Method | Links | |
|---|---|---|
| HDPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 85.79 | |
| HDPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 84.17 | |
| HDPOBackbone=Qwen3-4B, Number of samples=162026.06 | 83.26 | |
| Critique-GRPOBackbone=8B, Number of samples=162026.06 | 81.88 | |
| GRPOBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 80.92 | |
| LUFFYBackbone=Qwen-Math-7B, Number of samples=162026.06 | 79.68 | |
| GRPOBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 76.62 | |
| SimpleRLBackbone=Qwen-2.5-7B, Number of samples=162026.06 | 74.27 | |
| EDGE-GRPOBackbone=Qwen-7B, Number of samples=162026.06 | 73.79 | |
| OatBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 73.59 | |
| GRPOBackbone=Qwen3-4B, Number of samples=162026.06 | 73.32 | |
| Base ModelBackbone=Qwen2.5-Math-7B, Number of samples=162026.06 | 71.08 | |
| Base ModelBackbone=DS-R1-Llama-8B, Number of samples=162026.06 | 63.78 | |
| Base ModelBackbone=Qwen3-4B, Number of samples=162026.06 | 59.1 |