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

    Neuronpedia

    INDEX
    Explanations

    Separators

    np_max-act · gemini-2.0-flash

    This neuron detects long runs of punctuation characters (like repeated “=” or “–”) used as section-heading or separator lines.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Ок
    -0.07
     regain
    -0.07
     Для
    -0.07
    expect
    -0.06
    .dd
    -0.06
    ク
    -0.06
    }|
    -0.06
    'll
    -0.06
    oty
    -0.06
    ičky
    -0.06
    POSITIVE LOGITS
     Applies
    0.07
    iphone
    0.06
    derived
    0.06
    اعية
    0.06
    AttributeName
    0.06
    /release
    0.06
     HASH
    0.06
    σίας
    0.06
    ******↵↵
    0.06
     withholding
    0.06
    Act Density 0.002%

    No Known Activations