© 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. 129214
    Prev
    Next
    INDEX
    Explanations

    The neuron activates strongly on code-like tokens—identifiers, type names, numeric literals, and related symbols—i.e. fragments of programming or math notation.

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

    type, T, :, get

    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
    外部連結
    -1.12
     otros
    -1.07
    Bé
    -0.96
    dieran
    -0.95
     judiciales
    -0.95
     enviar
    -0.94
    zelfde
    -0.93
    zionato
    -0.92
    Câ
    -0.91
    grees
    -0.89
    POSITIVE LOGITS
    %;
    1.05
    _;
    1.04
    .*;
    1.00
     ';
    0.99
     honn
    0.98
    .';
    0.96
    ことです
    0.96
    It
    0.96
     européennes
    0.93
     inuti
    0.91
    Activations Density 0.178%

    No Known Activations