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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 25594
    Prev
    Next
    INDEX
    Explanations

    code structure delimiters

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

    The neuron fires on literal‐initializer tokens in code—especially empty or single‐character literals (e.g. “['',”, “([],”, “"",”, “'.',”) that mark empty strings, arrays, or similar literal defaults.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     یا
    -2.06
    ⁛
    -1.96
     เรา
    -1.95
     間取り
    -1.75
    に追加
    -1.73
    ſſen
    -1.72
     véhic
    -1.71
    我們
    -1.71
    Ahora
    -1.71
    čnega
    -1.69
    POSITIVE LOGITS
    きて
    1.92
     in
    1.84
    ‚
    1.80
     Перейти
    1.77
    re
    1.75
    鋱
    1.73
    д
    1.61
    が
    1.55
    itectura
    1.52
    ります
    1.52
    Activations Density 0.010%

    No Known Activations