© 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. 34-GEMMASCOPE-RES-131K
    7. 58888
    Prev
    Next
    INDEX
    Explanations

    code or structured data properties

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

    The neuron fires most strongly on numeric literals (digits and numbers) in code or JSON.

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

    commas used as separators after values in code, markup, and data structures.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    commas that separate items in JSON objects or function parameters.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     will
    -1.29
     should
    -1.23
     Despite
    -1.17
     for
    -1.16
     also
    -1.16
     still
    -1.13
     ostensibly
    -1.13
     not
    -1.13
     from
    -1.10
     despite
    -1.10
    POSITIVE LOGITS
     ™
    1.15
    诙
    1.12
     perche
    1.11
     mesmerizing
    1.10
    푆
    1.09
     momentous
    1.07
     shimmering
    1.07
     wondrous
    1.07
    摟
    1.06
     kirk
    1.05
    Activations Density 0.010%

    No Known Activations