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

    Neuronpedia

    INDEX
    Explanations

    Colon or quotes

    np_max-act-logits · gemini-2.0-flash

    colons and quotation marks that indicate dialogue or character speech in scripts and conversations.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jdhshshs138

    dialogue formatting, especially speaker name tags with colons, quoted speech, and bracketed stage directions.

    oai_token-act-pair · gpt-5Triggered by @jdhshshs138
    New Auto-Interp
    Negative Logits
     masturb
    -0.07
     comedian
    -0.07
    .Prot
    -0.07
     pcs
    -0.07
    重新
    -0.06
     Prä
    -0.06
     Lumia
    -0.06
     Fetish
    -0.06
     CGContext
    -0.06
     gorgeous
    -0.06
    POSITIVE LOGITS
    aset
    0.07
     finanzi
    0.07
    Entr
    0.07
    GRESS
    0.07
    视野
    0.07
     investors
    0.06
    arily
    0.06
    ulators
    0.06
    tection
    0.06
    anz
    0.06
    Act Density 0.057%

    No Known Activations