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

    Neuronpedia

    INDEX
    Explanations

    aki

    np_max-act · gemini-2.0-flash

    The neuron mainly detects substrings of proper nouns and other uncommon named-entity-like tokens (fragmented names and rare/foreign words).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
     Westminster
    -0.07
    щины
    -0.06
     richt
    -0.06
    _Account
    -0.06
    -stock
    -0.06
     Between
    -0.06
     transferring
    -0.06
     beware
    -0.06
     Beirut
    -0.06
    _es
    -0.06
    POSITIVE LOGITS
    .Room
    0.07
     öl
    0.07
    ">'.$
    0.07
    >,
    0.06
    .coordinate
    0.06
    .IsActive
    0.06
    BeginInit
    0.06
     조교
    0.06
    .dirname
    0.06
     todd
    0.06
    Act Density 5.138%

    No Known Activations