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 picking out numeric literal tokens—especially floating‐point numbers (decimals) in code.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    ROM
    -0.07
    	error
    -0.06
    Deferred
    -0.06
     bryster
    -0.06
     Secrets
    -0.06
     reimb
    -0.06
    懂
    -0.06
    abcdefghijklmnopqrstuvwxyz
    -0.06
    ]-$
    -0.06
    猛
    -0.06
    POSITIVE LOGITS
     indices
    0.07
     scop
    0.07
    шила
    0.07
    -shell
    0.06
     Vogue
    0.06
    ZZ
    0.06
    لام
    0.06
    58
    0.06
     але
    0.06
    _graphics
    0.06
    Act Density 0.009%

    No Known Activations