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

    Neuronpedia

    INDEX
    Explanations

    down

    np_max-act · gemini-2.0-flash

    The neuron flags small “thinking” tokens in first‐person introspective or desire statements (e.g. “deep,” “down,” “desire”)—it’s looking for inner reflections or personal wants.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     증
    -0.07
     nicknamed
    -0.06
     wget
    -0.06
    	byte
    -0.06
     香港
    -0.06
     середови
    -0.06
    (machine
    -0.06
     OPERATION
    -0.06
    �
    -0.06
    .ForegroundColor
    -0.06
    POSITIVE LOGITS
    ",-
    0.07
    erece
    0.07
    ,,,
    0.07
    _public
    0.07
    rift
    0.07
    idUser
    0.06
    ;-
    0.06
    QP
    0.06
     Junk
    0.06
     Compile
    0.06
    Act Density 0.009%

    No Known Activations