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

    Neuronpedia

    INDEX
    Explanations

    stop words

    np_max-act · gemini-2.0-flash

    The neuron is specialized to detect structural or metadata/control tokens (e.g. header delimiters and other special markers) rather than regular content.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    ]]=
    -0.07
    inces
    -0.07
    FETCH
    -0.07
     fast
    -0.07
     forfe
    -0.06
     feminist
    -0.06
     heures
    -0.06
    toLocale
    -0.06
     beef
    -0.06
     thirsty
    -0.06
    POSITIVE LOGITS
    INTERN
    0.06
    .Users
    0.06
    Moving
    0.06
    Keep
    0.06
    ея
    0.06
    ,(
    0.06
    .records
    0.06
    Length
    0.06
     dil
    0.06
    VIC
    0.06
    Act Density 0.104%

    No Known Activations