Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    MMLU benchmark, multitask understanding

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     Stoke
    0.52
    Если
    0.47
     Elite
    0.45
     Declare
    0.45
     Declar
    0.45
     Powers
    0.45
     APS
    0.44
     Use
    0.44
    APS
    0.43
     Excel
    0.43
    POSITIVE LOGITS
    hing
    0.45
    ithmet
    0.44
    breaking
    0.44
    лина
    0.44
     inability
    0.43
    ेस
    0.43
     alphan
    0.43
     unable
    0.42
    年間
    0.41
    cribes
    0.41
    Act Density 0.002%

    No Known Activations