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

    Neuronpedia

    INDEX
    Explanations

    Questions

    np_max-act · gemini-2.0-flash

    This neuron detects structural/control tokens marking conversation boundaries (end-of-turn/end-of-text and header/start markers).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
    setDefault
    -0.07
    Scrollbar
    -0.07
     explosion
    -0.07
    /xml
    -0.06
     становить
    -0.06
    оном
    -0.06
    くだ
    -0.06
     noir
    -0.06
     Stuff
    -0.06
    unload
    -0.06
    POSITIVE LOGITS
    ZE
    0.07
    ruptions
    0.06
    fin
    0.06
     روش
    0.06
    .*↵↵
    0.06
    ...,
    0.06
    agement
    0.06
     fertilizer
    0.06
    utch
    0.06
    (InputStream
    0.06
    Act Density 0.069%

    No Known Activations