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

    Neuronpedia

    INDEX
    Explanations

    "//"

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on the HTML/XML DOCTYPE and DTD declaration lines (e.g. the “<!DOCTYPE … PUBLIC …>” header).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    elements typical of HTML documents.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    енную
    -0.07
     utilizar
    -0.06
    arih
    -0.06
     In
    -0.06
    аного
    -0.06
    <std
    -0.06
    ична
    -0.06
    439
    -0.06
    woke
    -0.06
     Thermal
    -0.06
    POSITIVE LOGITS
    .reply
    0.06
    (milliseconds
    0.06
     -->↵↵
    0.06
     colossal
    0.06
    -threat
    0.06
    GI
    0.06
    selectorMethod
    0.06
     GDP
    0.06
    @Slf
    0.06
    __↵↵
    0.06
    Act Density 0.618%

    No Known Activations