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

    Neuronpedia

    INDEX
    Explanations

    comparisons indicating improvement or optimal choices

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating the best or optimal way to do something

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    naires
    -0.69
    mostly
    -0.67
    stairs
    -0.65
     Occasionally
    -0.63
    itially
    -0.62
    wick
    -0.60
     insofar
    -0.60
    adra
    -0.59
    ryu
    -0.59
     Mostly
    -0.57
    POSITIVE LOGITS
     than
    1.04
     Than
    0.92
     encaps
    0.78
    than
    0.76
     testament
    0.75
     illustration
    0.69
     exempl
    0.68
     nor
    0.67
     juxtap
    0.65
     deserving
    0.65
    Act Density 0.097%

    No Known Activations