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

    Neuronpedia

    INDEX
    Explanations

    code

    np_max-act · gemini-2.0-flash

    The neuron responds to tokens related to encoding or secret communication—words like “code,” “coded,” “cryptogram,” “secret,” and “information.”

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     mascot
    -0.06
     Besch
    -0.06
     zal
    -0.06
     Cum
    -0.06
    _OID
    -0.06
    _PROPERTY
    -0.06
    Outline
    -0.06
    ===========
    -0.06
    ated
    -0.06
     Messiah
    -0.06
    POSITIVE LOGITS
    �
    0.07
    ='',↵
    0.07
     Lands
    0.06
     Identified
    0.06
     international
    0.06
    ереч
    0.06
     Github
    0.06
    -token
    0.06
     completeness
    0.06
     northern
    0.06
    Act Density 0.035%

    No Known Activations