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

    Neuronpedia

    INDEX
    Explanations

    Repetitive actions

    np_max-act · gemini-2.0-flash

    The neuron fires on descriptive, rhythmic‐movement phrases introduced by “with each,” e.g. “with each step.”

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    (bytes
    -0.07
    _spaces
    -0.07
    Tick
    -0.07
     Fin
    -0.07
     nets
    -0.06
    _PAY
    -0.06
     metall
    -0.06
     яке
    -0.06
    eta
    -0.06
     sitcom
    -0.06
    POSITIVE LOGITS
     Ens
    0.06
    Ens
    0.06
    riage
    0.06
    <Void
    0.05
     sigmoid
    0.05
    atalog
    0.05
    braska
    0.05
     voks
    0.05
     sheriff
    0.05
    relu
    0.05
    Act Density 0.011%

    No Known Activations