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

    Neuronpedia

    INDEX
    Explanations

    phrases related to cleaning or removal, such as "washed away" and "swept under."

    oai_token-act-pair · gpt-3.5-turbo

    instances of cleaning or removal actions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    alter
    -0.81
    sit
    -0.75
    abet
    -0.74
    iola
    -0.72
    hold
    -0.71
    where
    -0.69
    ions
    -0.68
    MAT
    -0.66
    acion
    -0.66
    orial
    -0.65
    POSITIVE LOGITS
    nesday
    1.02
     ashore
    0.92
    ĸļ
    0.86
    ocument
    0.80
    adoes
    0.76
    uled
    0.73
    monton
    0.72
     aback
    0.71
     Yamato
    0.71
     destro
    0.70
    Act Density 0.058%

    No Known Activations