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

    Neuronpedia

    INDEX
    Explanations

    Conversational discussion

    np_max-act · gemini-2.0-flash

    The neuron fires on words that introduce or summarize spoken topics—especially verbs like “talk” or “discuss” (and the surrounding colons/dashes) that mark what’s being covered.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     electrodes
    -0.07
    资
    -0.07
     beat
    -0.06
    Unity
    -0.06
     intimate
    -0.06
     Christmas
    -0.06
     sperma
    -0.06
    _courses
    -0.06
     vice
    -0.06
    ية
    -0.06
    POSITIVE LOGITS
     LY
    0.06
    ||(
    0.06
    (memory
    0.06
     şer
    0.06
    [S
    0.06
    нив
    0.06
    {EIF
    0.06
     Thị
    0.06
    řil
    0.06
    /state
    0.06
    Act Density 0.019%

    No Known Activations