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

    Neuronpedia

    INDEX
    Explanations

    extremely negative descriptors related to violent actions or behaviors

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     soñ
    -0.64
     hadas
    -0.61
     houſe
    -0.61
    RTLR
    -0.60
     refugi
    -0.60
     administrativos
    -0.58
     normaux
    -0.58
     Chwiliwch
    -0.58
    unanje
    -0.58
     flattered
    -0.57
    POSITIVE LOGITS
     cruel
    1.15
     vicious
    1.10
     vile
    1.09
     depra
    1.06
     diabo
    1.05
     savage
    1.02
     evil
    1.02
     ruthless
    1.01
     brutal
    1.00
     sinister
    0.96
    Act Density 0.573%

    No Known Activations