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

    Neuronpedia

    INDEX
    Explanations

    ke

    np_max-act · gemini-2.0-flash

    The neuron fires on non-English (particularly Slavic) tokens—words with diacritics and letter combinations common in Czech/Slovak/Polish text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    phrases related to financial and bankruptcy terms.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8

    words with Latin-script diacritics (accented or special letters), especially from Central/Eastern European languages.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
     trous
    -0.08
     лит
    -0.07
    谋
    -0.07
    ovna
    -0.07
    chematic
    -0.07
     conspir
    -0.07
    เกษ
    -0.06
    _ioctl
    -0.06
    ustral
    -0.06
    [len
    -0.06
    POSITIVE LOGITS
    Morning
    0.07
     Saving
    0.06
     wav
    0.06
    !↵↵↵
    0.06
     scaffold
    0.06
    ाथ
    0.06
    Ask
    0.06
    pac
    0.06
     Starting
    0.06
    cripts
    0.06
    Act Density 0.379%

    No Known Activations