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

    Neuronpedia

    INDEX
    Explanations

    instances of the word "Almost" followed by an action or description

    oai_token-act-pair · gpt-3.5-turbo

    the word "almost" and similar qualifiers indicating approximation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     reserve
    -0.74
     specific
    -0.73
     repr
    -0.70
     other
    -0.69
     special
    -0.69
     mate
    -0.66
     corresponding
    -0.65
     particular
    -0.64
     prior
    -0.64
     right
    -0.63
    POSITIVE LOGITS
    Almost
    2.54
    Nearly
    2.35
     Nearly
    2.13
     Almost
    2.02
    almost
    1.77
    Quite
    1.30
    Most
    1.29
    Much
    1.28
    Half
    1.27
     Surprisingly
    1.24
    Act Density 0.014%

    No Known Activations