© 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. 104505
    Prev
    Next
    INDEX
    Explanations

    international words, fragments

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

    The neuron fires on substantive “content” tokens—i.e. longer, non‐stopword strings (often multi‐syllabic or non‐English words) rather than short function words or punctuation.

    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
     encuentras
    -2.05
    ichtete
    -1.99
    reurs
    -1.98
     Yorkers
    -1.93
    ühnen
    -1.80
    źdz
    -1.80
    when
    -1.76
    ilihan
    -1.67
    ers
    -1.67
    isable
    -1.66
    POSITIVE LOGITS
     are
    3.02
     themselves
    2.45
     to
    2.33
     THEIR
    2.25
    0
    2.19
     مختلف
    2.02
     THEY
    1.88
     škole
    1.88
     originalen
    1.84
    盦
    1.76
    Activations Density 0.080%

    No Known Activations