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

    Neuronpedia

    INDEX
    Explanations

    <

    np_max-act · gemini-2.0-flash

    The neuron is triggered by system‐level or roleplay metadata tokens and placeholders (e.g. “NAME_1,” header markers, instruction cues).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    enthusiastic, first-person-plural call-to-action lines in chat/roleplay dialogue.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
     Recorder
    -0.07
     Radeon
    -0.07
    .singletonList
    -0.06
    Receive
    -0.06
    -default
    -0.06
     sebeb
    -0.06
    _lead
    -0.06
     Panels
    -0.06
    .getSize
    -0.06
     kê
    -0.06
    POSITIVE LOGITS
    FULL
    0.07
    /menu
    0.06
    });↵
    0.06
    бор
    0.06
    彩
    0.06
    frag
    0.06
    алеж
    0.06
    ัด
    0.06
    ===========
    0.06
    !")↵
    0.06
    Act Density 1.970%

    No Known Activations