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

    Neuronpedia

    INDEX
    Explanations

    phrases related to criminal activities and legal processes

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to attempts and conspiracies to commit illegal actions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    felt
    -0.80
    ciating
    -0.79
    awed
    -0.73
    joy
    -0.67
     Printed
    -0.64
     Balanced
    -0.64
    thinking
    -0.64
    comings
    -0.63
    expected
    -0.62
     Chrys
    -0.62
    POSITIVE LOGITS
     assassinate
    1.54
     injure
    1.46
     violate
    1.44
     deceive
    1.43
     commit
    1.36
     steal
    1.35
     incite
    1.32
     smugg
    1.31
     overthrow
    1.31
     mislead
    1.28
    Act Density 0.181%

    No Known Activations