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

    Neuronpedia

    INDEX
    Explanations

    phrases describing difficult or challenging situations

    oai_token-act-pair · gpt-3.5-turbo

    the usage of conjunctions and phrases indicating conditional relationships

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    代
    -0.78
    PI
    -0.76
    911
    -0.72
    ãĥĨ
    -0.71
    UD
    -0.71
    Pen
    -0.71
    dayName
    -0.71
    ãĥĩãĤ£
    -0.70
    ãĤµ
    -0.70
    Paris
    -0.69
    POSITIVE LOGITS
     nonetheless
    1.01
     persists
    0.87
     persisted
    0.85
     nevertheless
    0.84
    etheless
    0.83
     prevailed
    0.79
     alas
    0.79
     emerges
    0.78
     curiously
    0.76
     retains
    0.72
    Act Density 0.285%

    No Known Activations