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

    Neuronpedia

    INDEX
    Explanations

    words that are misspelled or have typographical errors

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "err."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    creen
    -0.93
    esville
    -0.81
    eph
    -0.76
    ±
    -0.71
    ciating
    -0.70
    ķ
    -0.67
    vention
    -0.64
    Ŀ
    -0.61
    figure
    -0.61
    ļé
    -0.61
    POSITIVE LOGITS
    untled
    1.04
    antly
    0.97
    ands
    0.96
    rr
    0.96
    idge
    0.95
    utherford
    0.93
    abbit
    0.93
    uti
    0.90
    andom
    0.90
    ange
    0.88
    Act Density 0.041%

    No Known Activations