Scientific Reasoning on Material
81.5AccuracySRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| SRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 81.5 | |
| DRIFTBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.06 | 81.4 | |
| SC-SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 79.3 | |
| PGPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 78.7 | |
| SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 78.4 | |
| GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 77.1 | |
| DistILBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 76.2 | |
| SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.06 | 76.1 | |
| GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.06 | 73.5 | |
| Qwen3-8BBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.06 | 58.9 |