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

    Neuronpedia

    INDEX
    Explanations

    references to directions or locations within a building or house

    oai_token-act-pair · gpt-3.5-turbo

    references to the locations of "upstairs" and "downstairs."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    uality
    -0.75
    ck
    -0.74
    eming
    -0.67
    vg
    -0.66
    aceous
    -0.66
    nesota
    -0.66
    EVA
    -0.66
    UAL
    -0.65
    uated
    -0.63
    ually
    -0.62
    POSITIVE LOGITS
    stairs
    1.69
     downstairs
    1.34
     upstairs
    1.32
     stairs
    1.01
     challeng
    0.97
     mosqu
    0.90
     cellar
    0.86
     neighbour
    0.84
    MpServer
    0.82
     nodd
    0.82
    Act Density 0.006%

    No Known Activations