General Reasoning on GPQA (Average@4)
31.6Average@4JLB
Evaluation results
| Method | Links | |
|---|---|---|
| JLBBackbone=Qwen2.5-7B-Base, Algorithm=GRPO2025.12 | 31.6 | |
| RLPRBackbone=Qwen2.5-7B-Base, Algorithm=GRPO2025.12 | 31.3 | |
| LaTROBackbone=Qwen2.5-7B-Base, Algorithm=GRPO2025.12 | 31 | |
| CoVRLBackbone=Qwen2.5-7B-Base2025.12 | 30.4 | |
| RAVRBackbone=Qwen2.5-7B-Base, Algorithm=GRPO2025.12 | 30.2 | |
| VeriFreeBackbone=Qwen2.5-7B-Base, Algorithm=GRPO2025.12 | 28.9 | |
| Base ModelBackbone=Qwen2.5-7B-Base2025.12 | 26.1 |