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

    Neuronpedia

    INDEX
    Explanations

    source code

    np_max-act · gemini-2.0-flash

    This neuron detects Cyrillic-script (Russian) tokens — i.e., Russian-language text.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
    advert
    -0.06
    products
    -0.06
     presenter
    -0.06
    jug
    -0.06
     inadvert
    -0.06
    ')}}</
    -0.06
     Covers
    -0.06
     sing
    -0.06
    locals
    -0.06
     �
    -0.06
    POSITIVE LOGITS
    .TEXT
    0.06
     heed
    0.06
     Edwards
    0.06
     sophisticated
    0.06
    .dateFormat
    0.06
     €
    0.06
     philosophical
    0.06
    _players
    0.06
     Pasadena
    0.06
     preferably
    0.06
    Act Density 0.315%

    No Known Activations