List your company

Multiple-Choice Question Answering on MMLU STEM (test)

65.6Accuracy (ACC)DoLA
51.206454.943258.6862.4168Sep 28, 2025

Evaluation results

MethodLinks
DoLAModel=Qwen 2.5 7B2025.0965.687.810.714
KAPPAModel=Qwen 2.5 7B, Intervention Layers=62025.0965.5691.990.764
KAPPAModel=Qwen 2.5 7B, Intervention Layers=32025.0965.4592.70.724
Knowledge Probe (KP)Model=Qwen 2.5 7B2025.0965.361000
KAPPAModel=Qwen 2.5 7B, Intervention Layers=12025.0965.2493.170.721
CAAModel=Qwen 2.5 7B2025.0965.1989.50.708
BaseModel=Qwen 2.5 7B2025.0965.1789.70.703
CAAModel=Llama 3.1 8B2025.0954.8590.090.344
BaseModel=Llama 3.1 8B2025.0954.6591.120.321
KAPPAModel=Llama 3.1 8B, Intervention Layers=12025.0954.3490.560.684
KAPPAModel=Llama 3.1 8B, Intervention Layers=62025.0954.2384.580.899
KAPPAModel=Llama 3.1 8B, Intervention Layers=32025.0954.0386.30.755
DoLAModel=Llama 3.1 8B2025.0951.7671.360.533