© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 104281
    Prev
    Next
    INDEX
    Explanations

    Pendleton, Pendragon, Tren

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

    This neuron fires on four-letter alphabetic prefixes (e.g. “Pend-” in Pendleton/Pendragon/Pendle and “Tren-” in Trenitalia or “Quad-” in quadruple), effectively detecting those four-letter stems at the start of tokens.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    an
    -4.09
    a
    -3.77
    ”
    -3.38
    ly
    -2.88
    like
    -2.81
    st
    -2.69
    c
    -2.64
     –
    -2.59
    o
    -2.55
    T
    -2.52
    POSITIVE LOGITS
     selben
    2.61
    ركز
    2.42
    ﹡
    2.41
    𖧵
    2.38
    ishes
    2.34
    Ꭳ
    2.31
    ꯷
    2.31
    anyaan
    2.30
    𝙱
    2.28
    凌天
    2.25
    Activations Density 0.011%

    No Known Activations