Scientific Knowledge on GPQA Diamond
69.2AccuracySFT-S1.1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SFT-S1.1Model=Qwen3-4B2026.05 | 69.2 | 7.6 | |
| HINT TUNINGModel=Qwen3-4B2026.05 | 67.2 | 7.7 | |
| OriginalModel=Qwen3-4B2026.05 | 66.2 | 8.8 | |
| Budget PromptModel=Qwen3-4B2026.05 | 66.2 | 8.7 | |
| SFT-S1.1Model=DS-R1-32B2026.05 | 63.6 | 9.3 | |
| HINT TUNINGModel=DS-R1-32B2026.05 | 63.1 | 10.5 | |
| SFT-S1.1Model=DS-R1-14B2026.05 | 61.1 | 10 | |
| OriginalModel=DS-R1-14B2026.05 | 60.6 | 7.6 | |
| OriginalModel=DS-R1-32B2026.05 | 60.6 | 7.3 | |
| HINT TUNINGModel=DS-R1-14B2026.05 | 58.6 | 10.1 | |
| Budget PromptModel=DS-R1-32B2026.05 | 57.1 | 4.9 | |
| OriginalModel=DS-R1-7B2026.05 | 53.5 | 8.4 | |
| Budget PromptModel=DS-R1-14B2026.05 | 53 | 5.5 | |
| HINT TUNINGModel=DS-R1-7B2026.05 | 52 | 10.4 | |
| SFT-S1.1Model=DS-R1-7B2026.05 | 50.5 | 9.6 | |
| Budget PromptModel=DS-R1-7B2026.05 | 46.5 | 6.5 |