© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 130452
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on structural or formatting tokens in technical content—things like code‐ or markup‐syntax elements (e.g. punctuation in HTML tags, operators like “=”, angle brackets “< …>”, backslashes, math identifiers such as “dim”, “edge”, “concentration”, numeric literals, etc.).

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

    contract or code examples

    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
     YYYY
    -0.96
     mexico
    -0.96
     понимать
    -0.94
    stdbool
    -0.88
    Jahr
    -0.86
     espri
    -0.84
     carlo
    -0.84
    Therefore
    -0.84
    Categoría
    -0.84
     certe
    -0.84
    POSITIVE LOGITS
     vermute
    0.99
     „
    0.98
     wüns
    0.95
     dauern
    0.94
     каль
    0.93
    :[],
    0.92
     beobach
    0.92
     personalizado
    0.92
     gwaran
    0.91
    「
    0.91
    Activations Density 0.107%

    No Known Activations