© 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. 16093
    Prev
    Next
    INDEX
    Explanations

    code delimiters and structures

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

    The neuron selectively activates on JSON‐style structural elements—i.e. the property names (like “path,” “indent,” “newline,” “block”) and punctuation/comma tokens that appear in a serialized AST or JSON document.

    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
    Zubereitung
    -1.14
    č
    -1.09
    Descriere
    -1.08
     升
    -1.05
     起
    -1.05
    に対する
    -1.04
     Länder
    -1.02
    mais
    -1.02
    ссу
    -0.98
    ق
    -0.98
    POSITIVE LOGITS
     incríveis
    1.16
    分明
    1.14
     teléfonos
    1.14
     diamètre
    1.12
     accesorio
    1.11
     zainteres
    1.10
    になってしまう
    1.09
     merkwür
    1.09
     ejemplares
    1.09
     ogrom
    1.09
    Activations Density 0.025%

    No Known Activations