© 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. 129325
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on code tokens that are user-defined variable names (identifiers).

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

    code snippets and programming constructs

    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
     podoba
    -1.14
    kese
    -1.12
    menü
    -1.09
     berikan
    -1.09
     celebridades
    -1.05
    ynchron
    -1.05
    bunt
    -1.05
     akku
    -1.04
     projekty
    -1.02
    yeli
    -1.02
    POSITIVE LOGITS
     Then
    1.11
     was
    1.05
     then
    1.05
    Then
    0.99
    >
    0.98
    除此之外
    0.93
    Map
    0.92
    </b>
    0.91
    有一些
    0.91
    这也
    0.91
    Activations Density 0.001%

    No Known Activations