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

    Neuronpedia

    INDEX
    Explanations

    multilingual curiosities

    np_acts-logits-general · gemini-2.5-flash-lite

    list-style structure and outline formatting, especially numbered “top 10” items and section headings.

    oai_token-act-pair · gpt-5Triggered by @vetterc0

    The neuron is detecting tokens that represent numeric quantities—especially list‐size numbers or other multi‐digit numerals.

    oai_token-act-pair · o4-miniTriggered by @yooniel31
    New Auto-Interp
    Negative Logits
     TCM
    0.11
     ṣe
    0.11
    או
    0.11
    および
    0.11
     profiles
    0.11
     postup
    0.10
    5
    0.10
    MT
    0.10
    profiles
    0.10
    ও
    0.10
    POSITIVE LOGITS
    现象
    0.11
     infallible
    0.11
     言っ
    0.11
     ignorant
    0.10
     manures
    0.10
    ‍♂️
    0.10
     वाक्य
    0.10
     voisinage
    0.10
     esist
    0.10
    ยนต์
    0.10
    Act Density 18.376%

    No Known Activations