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

    Neuronpedia

    INDEX
    Explanations

    books and essays

    np_max-act · gemini-2.0-flash

    This neuron lights up on subword pieces containing accented characters (e.g. é, è, ô)—i.e. it detects fragments of Romance‐language words with diacritics.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     빌
    -0.06
     ANSW
    -0.06
     자유
    -0.06
     pizza
    -0.06
    ěř
    -0.06
    Speech
    -0.06
    Sw
    -0.06
    iz
    -0.06
     المع
    -0.06
     гем
    -0.06
    POSITIVE LOGITS
     #$
    0.07
    جه
    0.07
     [.
    0.06
     localObject
    0.06
    .FromSeconds
    0.06
    (optional
    0.06
    ;"><?
    0.06
     bourgeois
    0.06
    ]:=
    0.06
     scrimmage
    0.06
    Act Density 0.041%

    No Known Activations