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

    Neuronpedia

    INDEX
    Explanations

    instances of something failing

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "failed."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     accompan
    -0.77
    eous
    -0.76
    oooo
    -0.68
    ivan
    -0.68
     arrang
    -0.68
    pour
    -0.67
    laughs
    -0.66
    âĻ
    -0.66
    adays
    -0.65
    robe
    -0.63
    POSITIVE LOGITS
     failed
    3.55
    failed
    2.91
     Failed
    2.22
     fails
    2.16
     fail
    2.06
     failing
    2.00
     botched
    1.94
     failure
    1.92
     failures
    1.87
     unsuccessful
    1.81
    Act Density 0.015%

    No Known Activations