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

    Neuronpedia

    INDEX
    Explanations

    Code and logs

    np_max-act · gemini-2.0-flash

    This neuron detects occurrences of tool‐invocation keywords (like “run Search,” “run Calculator,” or “send Tweet”) in the instruction text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    .money
    -0.07
     Conf
    -0.07
    .urls
    -0.07
     gymn
    -0.07
    getTime
    -0.06
    tre
    -0.06
     CPI
    -0.06
    raquo
    -0.06
    .cart
    -0.06
    Display
    -0.06
    POSITIVE LOGITS
     грав
    0.06
    _excerpt
    0.06
    大阪
    0.06
     الأد
    0.06
    luví
    0.06
    \Events
    0.06
    .setBackground
    0.06
    0.06
     uvád
    0.06
     gol
    0.05
    Act Density 0.018%

    No Known Activations