Scientific Reasoning on GPQA-D (Acc@8)
46.46Acc@8GRPO-DBB
Evaluation results
| Method | Links | |
|---|---|---|
| GRPO-DBBBackbone=Qwen3-8B-Base, Training Dataset=DAPO-Math-17k2026.03 | 46.46 | |
| RePOBackbone=Qwen3-8B-Base, Training Dataset=DAPO-Math-17k2026.03 | 43.93 | |
| GRPOBackbone=Qwen3-8B-Base, Training Dataset=DAPO-Math-17k2026.03 | 42.93 | |
| GRPO-DBBBackbone=Qwen3-1.7B-Base, Training Dataset=DAPO-Math-17k2026.03 | 29.8 | |
| RePOBackbone=Qwen3-1.7B-Base, Training Dataset=DAPO-Math-17k2026.03 | 26.26 | |
| GRPOBackbone=Qwen3-1.7B-Base, Training Dataset=DAPO-Math-17k2026.03 | 20.7 |