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

    Neuronpedia

    INDEX
    Explanations

    us

    np_max-act · gemini-2.0-flash

    This neuron fires on direct imperative user instructions addressed to the assistant (e.g. “tell me,” “inform me,” “give me,” etc.), i.e. attempts to drive the assistant to break its normal rules.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     medication
    -0.07
    	Assert
    -0.07
    ゞ
    -0.06
     쇼
    -0.06
    人才
    -0.06
     prediction
    -0.06
    そんな
    -0.06
     Kam
    -0.06
    かった
    -0.06
    มหานคร
    -0.06
    POSITIVE LOGITS
    Skin
    0.08
    .Unlock
    0.07
     aute
    0.07
    ":"/
    0.06
     ense
    0.06
     gi
    0.06
     setSearch
    0.06
     saldo
    0.06
     FY
    0.06
    	old
    0.06
    Act Density 0.010%

    No Known Activations