Math Reasoning on SuperGPQA SGPQA-1k Math
46.5AccuracyDFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DFTModel=Qwen3 32B2025.10 | 46.5 | 46.5 | |
| VCOREModel=Qwen3 32B2025.10 | 45 | 45 | |
| OriginalModel=Qwen3 32B2025.10 | 43.9 | 43.9 | |
| RandomModel=Qwen3 32B2025.10 | 40.3 | 40.3 | |
| SFTModel=Qwen3 32B2025.10 | 39.8 | 39.8 | |
| iw-SFTModel=Qwen3 32B2025.10 | 39.4 | 39.4 | |
| DFTModel=Qwen3 8B2025.10 | 37.9 | 37.9 | |
| VCOREModel=Qwen3 8B2025.10 | 36.8 | 36.8 | |
| RandomModel=Qwen3 8B2025.10 | 36.2 | 36.2 | |
| iw-SFTModel=Qwen3 8B2025.10 | 34.6 | 34.6 | |
| VCOREModel=Qwen3 4B2025.10 | 34.3 | 34.3 | |
| OriginalModel=Qwen3 8B2025.10 | 34.3 | 34.3 | |
| SFTModel=Qwen3 8B2025.10 | 34 | 34 | |
| DFTModel=Qwen3 4B2025.10 | 32.4 | 32.4 | |
| iw-SFTModel=Qwen3 4B2025.10 | 31 | 31 | |
| SFTModel=Qwen3 4B2025.10 | 30.5 | 30.5 | |
| RandomModel=Qwen3 4B2025.10 | 30.5 | 30.5 | |
| OriginalModel=Qwen3 4B2025.10 | 29.1 | 29.1 | |
| OriginalModel=LLaMA3.1 8B Instruct2025.10 | 18.6 | 18.6 | |
| RandomModel=LLaMA3.1 8B Instruct2025.10 | 15.1 | 15.1 | |
| DFTModel=LLaMA3.1 8B Instruct2025.10 | 13.9 | 13.9 | |
| VCOREModel=LLaMA3.1 8B Instruct2025.10 | 12.8 | 12.8 | |
| SFTModel=LLaMA3.1 8B Instruct2025.10 | 12.1 | 12.1 | |
| iw-SFTModel=LLaMA3.1 8B Instruct2025.10 | 11.6 | 11.6 |