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

    Neuronpedia

    INDEX
    Explanations

    repetitive phrases, especially variations of the word "said."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     Morty
    -0.65
    REL
    -0.58
     voed
    -0.58
     Jum
    -0.57
    Sk
    -0.56
    ofern
    -0.56
    </table>
    -0.56
     Flä
    -0.56
     Columb
    -0.55
     denomin
    -0.55
    POSITIVE LOGITS
     said
    3.12
    said
    2.74
    Said
    2.65
     Said
    2.62
     SAID
    2.58
     says
    2.05
     Says
    1.97
    Says
    1.88
    says
    1.88
     SAYS
    1.85
    Act Density 0.085%

    No Known Activations