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

    Neuronpedia

    INDEX
    Explanations

    contractions formed with "don't"

    oai_token-act-pair · gpt-3.5-turbo

    negations and contractions related to disbelief or failure to act

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    Reader
    -0.70
    ãĤ¶
    -0.69
    ãĤ¦ãĤ¹
    -0.68
    OST
    -0.68
    çĶŁ
    -0.66
    Fra
    -0.64
    İ
    -0.63
    ONSORED
    -0.63
    NetMessage
    -0.62
    ergus
    -0.60
    POSITIVE LOGITS
     comply
    1.11
     agree
    1.10
     succeed
    1.06
     cooperate
    1.04
     want
    1.03
     heed
    1.02
     adhere
    0.97
     wanna
    0.97
     obey
    0.92
     hurry
    0.92
    Act Density 0.098%

    No Known Activations