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

    Neuronpedia

    INDEX
    Explanations

    completing sentences/quoting

    np_max-act · gemini-2.0-flash

    The neuron is detecting special control tokens that mark the boundaries and headers of conversation segments (e.g. “<|start_header_id|>”, end-of-turn markers, and other segment delimiters).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    EW
    -0.07
    INESS
    -0.06
    โช
    -0.06
    별
    -0.06
    	friend
    -0.06
    _ant
    -0.06
     کوت
    -0.06
    aml
    -0.06
    .ContextCompat
    -0.06
    AutoresizingMaskIntoConstraints
    -0.06
    POSITIVE LOGITS
    ýval
    0.07
    =head
    0.07
     Suppose
    0.06
    (arc
    0.06
     فاصله
    0.06
     Tk
    0.06
     생각
    0.06
     Continue
    0.06
     kullanım
    0.06
     egreg
    0.06
    Act Density 0.033%

    No Known Activations