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

    Neuronpedia

    INDEX
    Explanations

    specific instances where understanding or knowledge is explicitly mentioned

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "understands."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    onies
    -0.86
    ads
    -0.74
     downed
    -0.73
    collection
    -0.72
    cial
    -0.67
    haps
    -0.65
    pmwiki
    -0.64
    ammy
    -0.63
    rav
    -0.63
    ogue
    -0.63
    POSITIVE LOGITS
     understands
    0.96
    ¿½
    0.89
     understood
    0.80
     Understand
    0.78
    ãĤ©
    0.78
     SERV
    0.77
    ledged
    0.73
    terday
    0.72
     Trident
    0.70
    uitive
    0.69
    Act Density 0.003%

    No Known Activations