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

    Neuronpedia

    INDEX
    Explanations

    Code/data

    np_max-act · gemini-2.0-flash

    The neuron spikes on long runs of the same (or very similar) tokens in a row—that is, repeated sequences of identical or nearly identical words.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     melt
    -0.08
     extensive
    -0.07
     STATES
    -0.06
    Uri
    -0.06
     문자
    -0.06
     زر
    -0.06
    phrase
    -0.06
    gzip
    -0.06
    	arg
    -0.06
    _other
    -0.06
    POSITIVE LOGITS
     ویژه
    0.08
    Ù
    0.07
     discoveries
    0.07
     ра
    0.07
     erotiske
    0.07
    ."',
    0.06
    _attachment
    0.06
     arsenal
    0.06
     MongoClient
    0.06
     playlists
    0.06
    Act Density 0.003%

    No Known Activations