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

    Neuronpedia

    INDEX
    Explanations

    letter probability calculations

    np_max-act · gemini-2.0-flash

    The neuron activates on pairs of identical letters—i.e. double‐letter sequences like “nn,” “rr,” “pp,” etc.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Datagram
    -0.08
    _contr
    -0.07
     Ad
    -0.07
     Lyft
    -0.06
    536
    -0.06
    ,"\
    -0.06
    Ta
    -0.06
     limiting
    -0.06
    .spacing
    -0.06
     Lag
    -0.06
    POSITIVE LOGITS
     يجب
    0.06
     Soviets
    0.06
     whole
    0.06
     확인
    0.06
    บาย
    0.06
    ;';↵
    0.06
    خصوص
    0.06
    аю
    0.06
    _height
    0.06
    dfunding
    0.06
    Act Density 0.004%

    No Known Activations