© 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. 22-GEMMASCOPE-RES-131K
    7. 131044
    Prev
    Next
    INDEX
    Explanations

    The neuron fires strongly on “structural” code‐or‐doc tokens – things like section‐separator lines of dashes, package or domain names (e.g. imports, GitHub or Eclipse URLs), and other dotted/qualified identifiers.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    code and comments

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     MÁ
    -1.12
    cím
    -1.01
    fertigen
    -0.96
     graisse
    -0.95
     arranges
    -0.93
    Má
    -0.93
    OPPO
    -0.93
     neemt
    -0.92
     bakteri
    -0.91
    Mening
    -0.90
    POSITIVE LOGITS
    lelő
    0.93
     bustling
    0.91
    /*",
    0.90
     bezahlt
    0.90
     dostęp
    0.88
    acular
    0.87
     welchen
    0.86
    CH
    0.85
     tatuagem
    0.85
     their
    0.84
    Activations Density 0.019%

    No Known Activations