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

    Neuronpedia

    INDEX
    Explanations

    words related to actions or events in specific contexts, such as "get out," "see," "walked in," "go underwater," "gets back into," and "open."

    oai_token-act-pair · gpt-3.5-turbo

    phrases that indicate actions or occurrences related to getting, seeing, or moving

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    gart
    -0.70
    mong
    -0.68
    WER
    -0.68
     aver
    -0.67
    oÄŁ
    -0.66
    avery
    -0.65
    ape
    -0.65
    uther
    -0.62
    never
    -0.62
    ono
    -0.62
    POSITIVE LOGITS
     Stamford
    0.65
     Hitman
    0.65
    issan
    0.62
     Chimera
    0.61
     Shanghai
    0.60
     landfall
    0.60
     PTS
    0.58
     Lanc
    0.58
     Ju
    0.57
    Reviewer
    0.56
    Act Density 0.250%

    No Known Activations