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

    Neuronpedia

    INDEX
    Explanations

    PRODUCT

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on uppercase, underscore-separated placeholder names (e.g. PRODUCT_NAME, EXECUTABLE_NAME) in build or plist variable expressions.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    `
↵
    -0.07
    :")↵
    -0.07
    belt
    -0.06
    ≤
    -0.06
     هذا
    -0.06
    iotic
    -0.06
     day
    -0.06
    Egypt
    -0.06
     हम
    -0.06
     forgotten
    -0.06
    POSITIVE LOGITS
     曰
    0.06
    bír
    0.06
     succesfully
    0.06
     TInt
    0.06
    ，不过
    0.06
    �情
    0.06
    Debe
    0.06
     Cumhuriyeti
    0.06
     reciprocal
    0.05
    _TRA
    0.05
    Act Density 0.002%

    No Known Activations