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

    Neuronpedia

    INDEX
    Explanations

    I'm sorry, but based on the provided activations, it is difficult to clearly summarize what Neuron 4 is looking for. There is a variety of seemingly random text content, making it challenging to pinpoint a specific pattern or theme that the neuron is primarily activating for

    oai_token-act-pair · gpt-3.5-turbo

    mentions of the website "Infowars."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
     Rolls
    -0.68
     bye
    -0.66
     Dunham
    -0.62
     uninterrupted
    -0.62
     bom
    -0.61
     valiant
    -0.59
     Winston
    -0.58
     successors
    -0.58
     davidjl
    -0.57
     Hug
    -0.57
    POSITIVE LOGITS
    owship
    1.01
    cial
    0.99
    ructure
    0.97
    iband
    0.97
    renheit
    0.95
    essim
    0.90
    ruct
    0.89
    urated
    0.81
    aneous
    0.79
    luster
    0.79
    Act Density 0.036%

    No Known Activations