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

    Neuronpedia

    INDEX
    Explanations

    Start of sentences

    np_max-act · gemini-2.0-flash

    The neuron mainly activates on numeric date tokens—especially four-digit years (often in parentheses) indicating founding or event dates.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    蓝
    -0.07
     данных
    -0.07
    .StObject
    -0.07
     rejecting
    -0.06
     Blank
    -0.06
    झ
    -0.06
     wd
    -0.06
     Patrick
    -0.06
     feedback
    -0.06
     Vine
    -0.06
    POSITIVE LOGITS
    мат
    0.07
    ;p
    0.06
    deps
    0.06
    ','=',$
    0.06
    :any
    0.06
    ="">↵
    0.06
    <head
    0.06
     ISP
    0.06
    _DP
    0.06
     '%$
    0.06
    Act Density 0.239%

    No Known Activations