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

    Neuronpedia

    INDEX
    Explanations

    dates in the format of year/month/day

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Negative Logits
     OU
    -0.80
     Brawl
    -0.77
     Thrones
    -0.76
     Bound
    -0.75
     Throne
    -0.75
     Talent
    -0.74
     Fisheries
    -0.74
     Baal
    -0.74
     denomination
    -0.73
     plaque
    -0.73
    POSITIVE LOGITS
    why
    1.56
    hillary
    1.40
    exclusive
    1.38
    how
    1.36
    maxwell
    1.31
    another
    1.29
    south
    1.29
    what
    1.29
    leon
    1.28
    john
    1.25
    Act Density 0.031%

    No Known Activations