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

    Neuronpedia

    INDEX
    Explanations

    contrast and emphasis

    np_max-act · gemini-2.0-flash

    The neuron detects emphasis words—comparative or superlative modifiers and intensifiers (e.g. “most,” “biggest,” “far,” “by far”).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Chang
    -0.07
    ATE
    -0.07
    ابط
    -0.07
    ियल
    -0.07
    -middle
    -0.07
    وات
    -0.07
     Gree
    -0.07
    gate
    -0.06
    -0.06
     semiconductor
    -0.06
    POSITIVE LOGITS
    /LICENSE
    0.07
    Capacity
    0.06
    @yahoo
    0.06
     ارز
    0.06
     elephant
    0.06
     orderId
    0.06
    ValueGenerationStrategy
    0.06
    [:]
    0.06
    _VEC
    0.06
    _product
    0.06
    Act Density 0.053%

    No Known Activations