Scientific Reasoning on Chemistry (accuracy)
83AccuracySRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| SRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 83 | |
| DRIFTBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.06 | 82 | |
| SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 80.9 | |
| DistILBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 80.8 | |
| SC-SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 80.6 | |
| SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.06 | 78.9 | |
| PGPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 77.6 | |
| GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.06 | 74.5 | |
| GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.06 | 65.6 | |
| Qwen3-8BBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.06 | 41.2 |