Reasoning on DAPO
55.7Pass@1GRPO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GRPOBackbone=Qwen3-8B-Base, Supervision Type=Labeled reward data2026.06 | 55.7 | 81.5 | |
| GRPOBackbone=Qwen3-4B-Base, Supervision Type=Labeled reward data2026.06 | 49.9 | 77 | |
| TTRL-CoCoVBackbone=Qwen3-8B-Base, Supervision Type=Label-free2026.06 | 45.3 | 77.6 | |
| Co-rewardingBackbone=Qwen3-8B-Base, Supervision Type=Label-free2026.06 | 44.7 | 63.9 | |
| TTRL-CoCoVBackbone=Qwen3-4B-Base, Supervision Type=Label-free2026.06 | 43 | 76 | |
| Co-rewardingBackbone=Qwen3-4B-Base, Supervision Type=Label-free2026.06 | 40.1 | 60.1 | |
| TTRLBackbone=Qwen3-8B-Base, Supervision Type=Label-free2026.06 | 38.7 | 64.3 | |
| SCRLBackbone=Qwen3-8B-Base, Supervision Type=Label-free2026.06 | 38.4 | 63 | |
| SCRLBackbone=Qwen3-4B-Base, Supervision Type=Label-free2026.06 | 34.7 | 60.6 | |
| TTRLBackbone=Qwen3-4B-Base, Supervision Type=Label-free2026.06 | 31.2 | 55.1 | |
| BaseBackbone=Qwen3-8B-Base, Supervision Type=Label-free2026.06 | 29.3 | 69.9 | |
| BaseBackbone=Qwen3-4B-Base, Supervision Type=Label-free2026.06 | 28.7 | 65.5 |