List your company

Multiple-Choice Question Answering on MMLU Humanities (test)

62.47Accuracy (Humanities)Knowledge Probe (KP)
55.824457.549759.27561.0003Sep 28, 2025

Evaluation results

MethodLinks
Knowledge Probe (KP)Model=Qwen 2.5 7B2025.0962.471000
KAPPAModel=Qwen 2.5 7B, Intervention Layers=62025.0961.9285.220.703
KAPPAModel=Qwen 2.5 7B, Intervention Layers=32025.0961.0482.230.72
KAPPAModel=Qwen 2.5 7B, Intervention Layers=12025.0960.7782.030.751
KAPPAModel=Llama 3.1 8B, Intervention Layers=62025.0960.73870.446
KAPPAModel=Llama 3.1 8B, Intervention Layers=12025.0960.6787.150.533
KAPPAModel=Llama 3.1 8B, Intervention Layers=32025.0960.3285.190.456
CAAModel=Qwen 2.5 7B2025.0959.9278.420.777
BaseModel=Qwen 2.5 7B2025.0959.9178.50.777
DoLAModel=Qwen 2.5 7B2025.0959.8578.640.811
CAAModel=Llama 3.1 8B2025.0959.6277.560.487
BaseModel=Llama 3.1 8B2025.0958.5777.940.469
DoLAModel=Llama 3.1 8B2025.0956.0869.350.504