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

    Neuronpedia

    INDEX
    Explanations

    Over

    np_max-act · gemini-2.0-flash

    The neuron responds to tokens that mark the start of a new broadcast segment or speaker turn—particularly the capitalized words and short phrases used as show intros or host cues.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     thực
    -0.07
    (download
    -0.07
     Ripple
    -0.07
    &q
    -0.07
     pane
    -0.07
    .pr
    -0.06
    Appointment
    -0.06
    	bt
    -0.06
    .getItems
    -0.06
     Venue
    -0.06
    POSITIVE LOGITS
     підвищ
    0.07
     зобов
    0.07
     né
    0.07
    esy
    0.06
     спіл
    0.06
    .MULT
    0.06
    .proxy
    0.06
    existence
    0.06
    】↵↵
    0.06
    (FLAGS
    0.06
    Act Density 0.014%

    No Known Activations