© 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. 129087
    Prev
    Next
    INDEX
    Explanations

    The neuron selectively activates on all-caps programming tokens (e.g. preprocessor macros and constant identifiers).

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

    C++ development and programming terms

    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
     from
    -1.60
    為に
    -1.13
    に行った
    -1.12
     limitée
    -1.08
     belga
    -1.07
    ようです
    -1.04
    __':
    -1.02
    ůž
    -1.02
     told
    -1.02
    ようだ
    -1.00
    POSITIVE LOGITS
    で
    1.21
    では
    1.15
    新潟
    1.13
    は
    1.11
    かっ
    1.07
    醍
    1.06
     ModelRenderer
    1.06
    <bos>
    1.05
    がありますが
    1.05
     Grüßen
    1.04
    Activations Density 0.004%

    No Known Activations