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

    Neuronpedia

    INDEX
    Explanations

    instances of attempted actions or interventions

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "tried"

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    scribe
    -0.77
    head
    -0.76
    inant
    -0.67
    cedented
    -0.66
    Quality
    -0.66
    ificantly
    -0.65
    Production
    -0.64
    thus
    -0.63
    hi
    -0.62
    requisite
    -0.62
    POSITIVE LOGITS
     unsuccessfully
    1.36
     valiant
    0.81
     desperately
    0.81
     harder
    0.78
     vain
    0.72
    nces
    0.72
     repeatedly
    0.71
     tried
    0.69
     Pett
    0.69
    nesday
    0.69
    Act Density 0.049%

    No Known Activations