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

    Neuronpedia

    INDEX
    Explanations

    HTML and forum formatting

    np_max-act · gemini-2.0-flash

    tokens that mark metadata or conversation boundaries (special control tokens like end-of-header, end-of-text, and similar role/segment markers).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
    /owl
    -0.07
    STREAM
    -0.07
    OTION
    -0.07
     rounds
    -0.07
    ­
    -0.07
    具
    -0.07
     fila
    -0.06
    cheng
    -0.06
     defining
    -0.06
    /photo
    -0.06
    POSITIVE LOGITS
     username
    0.07
     orderId
    0.06
     `(
    0.06
     DataView
    0.06
     سر
    0.06
    .dataset
    0.06
     bitterly
    0.06
     ObjectOutputStream
    0.06
    .identifier
    0.06
     شن
    0.06
    Act Density 0.320%

    No Known Activations