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

    Neuronpedia

    INDEX
    Explanations

    numbers

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on numeric tokens—especially digits and ordinals used as list markers (e.g. “1,” “2,” “first,” “second,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     şaş
    -0.07
     portal
    -0.06
     wget
    -0.06
     dma
    -0.06
    fffffff
    -0.06
     consortium
    -0.06
     provisioning
    -0.06
     legitimately
    -0.06
     fout
    -0.06
     hypoth
    -0.06
    POSITIVE LOGITS
    (att
    0.06
    bin
    0.06
    /response
    0.06
    الية
    0.06
     الولايات
    0.06
    reshape
    0.06
    레이
    0.06
    YP
    0.06
    	LL
    0.06
    UGH
    0.06
    Act Density 0.069%

    No Known Activations