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

    Neuronpedia

    INDEX
    Explanations

    Japanese honorifics

    np_max-act · gemini-2.0-flash

    The neuron activates on Japanese polite honorific prefixes (the “お” and “ご” often used to show respect).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Kenn
    -0.08
    Better
    -0.07
    売
    -0.07
     murderous
    -0.07
    survey
    -0.07
    _Show
    -0.06
    _upgrade
    -0.06
    Nh
    -0.06
     dung
    -0.06
    marshall
    -0.06
    POSITIVE LOGITS
    _stock
    0.07
     otom
    0.07
    하신
    0.06
    .custom
    0.06
    하시
    0.06
     �
    0.06
     (![
    0.06
     výši
    0.06
     zip
    0.06
    ::::::::::::::::::::::::::::::::
    0.06
    Act Density 0.009%

    No Known Activations