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

    Neuronpedia

    INDEX
    Explanations

    Code logos

    np_max-act · gemini-2.0-flash

    The neuron activates on runs of underscores (and similar repetitive punctuation), i.e. the horizontal line elements used in ASCII art.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    �
    -0.07
    ARB
    -0.07
    علی
    -0.06
     پخش
    -0.06
     فراهم
    -0.06
     숨
    -0.06
    歴
    -0.06
    ується
    -0.06
     varsa
    -0.06
     değildir
    -0.06
    POSITIVE LOGITS
     ______
    0.07
    ____
    0.07
     ___
    0.07
     Appro
    0.07
    approve
    0.07
    ___
    0.06
     Transitional
    0.06
     _____
    0.06
     neoliberal
    0.06
     Buckley
    0.06
    Act Density 0.003%

    No Known Activations