Science Reasoning on GPQA (Avg@4)
39.4Avg@4DARL
Evaluation Results
| Method | Links | |
|---|---|---|
| DARLBase=Base, Verifier=None2026.01 | 39.4 | |
| Oat-ZeroBase=Math, Verifier=Rule2026.01 | 38.8 | |
| SimpleRL-ZooBase=Math, Verifier=Rule2026.01 | 38.4 | |
| RLPRBase=Base, Verifier=None2026.01 | 37.6 | |
| General ReasonerBase=Base, Verifier=Model2026.01 | 37.4 | |
| DARLBase=Inst, Verifier=None2026.01 | 36.9 | |
| VeriFreeBase=Base, Verifier=None2026.01 | 36.7 | |
| RLPRBase=Inst, Verifier=None2026.01 | 36.5 | |
| SimpleRL-ZooBase=Base, Verifier=Rule2026.01 | 36.2 | |
| RLVRBase=Base, Verifier=Rule2026.01 | 36.2 | |
| RLVRBase=Inst, Verifier=Rule2026.01 | 36 | |
| Qwen2.5-7B-InstBase=-, Verifier=None2026.01 | 34.2 | |
| TTRLBase=Base, Verifier=Rule2026.01 | 34.1 | |
| Qwen2.5-7BBase=-, Verifier=None2026.01 | 32.4 | |
| PRIMEBase=Math, Verifier=Rule2026.01 | 32.1 | |
| Llama3.1-8B-InstBase=-, Verifier=None2026.01 | 31.6 |