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

    Neuronpedia

    INDEX
    Explanations

    phrases related to proper handling or adherence to procedures

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "properly" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    =-=-=-=-=-=-=-=-
    -0.86
     Sark
    -0.66
     Memories
    -0.66
     Spice
    -0.65
     Mehran
    -0.65
    tery
    -0.64
     Obi
    -0.62
    RY
    -0.62
     Tao
    -0.62
     Sketch
    -0.61
    POSITIVE LOGITS
     properly
    1.07
     behaved
    1.02
     exting
    1.02
     adequately
    0.96
     assimil
    0.89
     fitted
    0.89
     alloc
    0.88
     equipped
    0.88
     initialized
    0.86
     initialize
    0.86
    Act Density 0.006%

    No Known Activations