List your company

Multiple-Choice Question Answering on Mechanistic Interpretability Benchmark (MIB) MCQA (standard)

0.04CMDEAP-IG-inputs
0.03640.06070.0850.1093Feb 10, 2026

Evaluation results

MethodLinks
EAP-IG-inputsModel=OPT-1.3B2026.020.040.96
EAP-IG-inputsModel=Qwen2.5-0.5B2026.020.0595
EAP-IG-inputsModel=Llama3.2-1B2026.020.0595
EAPModel=OPT-1.3B2026.020.050.95
EAPModel=Qwen2.5-0.5B2026.020.0694
Circuit FingerprintModel=OPT-1.3B2026.020.070.93
Circuit FingerprintModel=Qwen2.5-0.5B2026.020.0992
EAPModel=Llama3.2-1B2026.020.130.87
Circuit FingerprintModel=Llama3.2-1B2026.020.130.87