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

    Neuronpedia

    INDEX
    Explanations

    dialogue snippets

    np_max-act · gemini-2.0-flash

    The neuron detects polite address or apology phrases (e.g. “Madam,” “I’m sorry,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    -0.07
     pow
    -0.07
    .inverse
    -0.07
    find
    -0.07
    ний
    -0.06
    iera
    -0.06
     jew
    -0.06
    ween
    -0.06
     например
    -0.06
    culated
    -0.06
    POSITIVE LOGITS
    (/^\
    0.07
    _FLUSH
    0.07
     вступ
    0.06
    BASH
    0.06
     testim
    0.06
    едак
    0.06
    capture
    0.06
    .DELETE
    0.06
     uncont
    0.06
     WaitForSeconds
    0.06
    Act Density 0.011%

    No Known Activations