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

    Neuronpedia

    INDEX
    Explanations

    12 without zeros filtered out, the main thing this neuron does is find numbers as potential references or annotations within the text

    oai_token-act-pair · gpt-3.5-turbo

    references to specific numerical values or identifiers, particularly focusing on the number 12

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     tremend
    -0.83
     behavi
    -0.82
    chwitz
    -0.80
    merce
    -0.80
    ileaks
    -0.78
    ifully
    -0.77
    icist
    -0.77
    iciary
    -0.77
    ierrez
    -0.76
    Ô
    -0.76
    POSITIVE LOGITS
    91
    1.01
    34
    1.01
    650
    0.98
    92
    0.98
    87
    0.97
    82
    0.94
    71
    0.93
    76
    0.93
    02
    0.93
    94
    0.93
    Act Density 0.037%

    No Known Activations