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

    Neuronpedia

    INDEX
    Explanations

    Filenames, internet content

    np_max-act · gemini-2.0-flash

    The neuron fires on isolated metadata‐style tokens (e.g. attribute names, file‐related keywords, tags/flags, and other structured identifiers).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     잡
    -0.07
     Castro
    -0.06
    еві
    -0.06
     Chương
    -0.06
    映
    -0.06
    ENTA
    -0.06
    ाइस
    -0.06
     От
    -0.06
    illos
    -0.06
     очі
    -0.06
    POSITIVE LOGITS
     %%↵
    0.07
     hues
    0.07
    .cvtColor
    0.07
    'use
    0.06
     UIButton
    0.06
     noexcept
    0.06
     gql
    0.06
     yog
    0.06
     clicking
    0.06
    ],[
    0.06
    Act Density 0.123%

    No Known Activations