List your company

Code Reasoning on SuperGPQA Code SGPQA-1k (Accuracy, Pass@1)

47.4AccuracyDFT
3.40814.82926.2537.671Oct 31, 2025

Evaluation results

MethodLinks
DFTModel=Qwen3 32B2025.1047.447.4
RandomModel=Qwen3 32B2025.1045.545.5
OriginalModel=Qwen3 32B2025.1043.943.9
VCOREModel=Qwen3 32B2025.1043.143.1
iw-SFTModel=Qwen3 32B2025.1042.442.4
SFTModel=Qwen3 32B2025.1040.640.6
SFTModel=Qwen3 8B2025.1037.937.9
DFTModel=Qwen3 8B2025.1037.537.5
iw-SFTModel=Qwen3 8B2025.1035.935.9
RandomModel=Qwen3 8B2025.1035.935.9
VCOREModel=Qwen3 8B2025.1035.435.4
OriginalModel=Qwen3 8B2025.1034.334.3
DFTModel=Qwen3 4B2025.1033.433.4
iw-SFTModel=Qwen3 4B2025.1032.632.6
VCOREModel=Qwen3 4B2025.1032.332.3
SFTModel=Qwen3 4B2025.1031.831.8
RandomModel=Qwen3 4B2025.1031.331.3
OriginalModel=Qwen3 4B2025.1029.129.1
OriginalModel=LLaMA3.1 8B Instruct2025.1018.618.6
VCOREModel=LLaMA3.1 8B Instruct2025.1015.815.8
SFTModel=LLaMA3.1 8B Instruct2025.109.59.5
RandomModel=LLaMA3.1 8B Instruct2025.108.58.5
iw-SFTModel=LLaMA3.1 8B Instruct2025.106.66.6
DFTModel=LLaMA3.1 8B Instruct2025.105.15.1