Reasoning on GPQA Diamond Expert-level OOD
37.3pass@1LUFFY
Evaluation Results
| Method | Links | |
|---|---|---|
| LUFFYBackbone=Qwen2.5-Math-7B2025.10 | 37.3 | |
| Scaf-GRPOBackbone=Qwen2.5-Math-7B2025.10 | 37.3 | |
| Scaf-GRPOBackbone=Qwen2.5-7B-Base2025.10 | 35.8 | |
| LUFFYBackbone=Qwen2.5-7B-Base2025.10 | 34.4 | |
| SimpleRL-ZeroBackbone=Qwen2.5-Math-7B2025.10 | 33.3 | |
| Oat-ZeroBackbone=Qwen2.5-Math-7B2025.10 | 33.3 | |
| Vanilla GRPOBackbone=Qwen2.5-7B-Base2025.10 | 33.3 | |
| Vanilla GRPOBackbone=Qwen2.5-Math-7B2025.10 | 32.3 | |
| Base ModelBackbone=Qwen2.5-Math-7B2025.10 | 24.7 |