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

    Neuronpedia

    INDEX
    Explanations

    Code/data

    np_max-act · gemini-2.0-flash

    This neuron fires on programming‐style tokens—things like code punctuation, braces, keyword identifiers, hyphenated attribute or function names, URLs, backticks, etc.—i.e. elements of code snippets or markup.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    _echo
    -0.08
    Coupon
    -0.06
     добре
    -0.06
    гл
    -0.06
    ompiler
    -0.06
    로서
    -0.06
    ngör
    -0.06
     DISTINCT
    -0.06
     Museum
    -0.06
     здоб
    -0.06
    POSITIVE LOGITS
     laughs
    0.08
     Dana
    0.07
    ahan
    0.07
     pubb
    0.07
    >]
    0.06
     learn
    0.06
     Delta
    0.06
    aired
    0.06
     Contains
    0.06
     cres
    0.06
    Act Density 0.427%

    No Known Activations