© 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. 10-GEMMASCOPE-RES-131K
    7. 38935
    Prev
    Next
    INDEX
    Explanations

    require statements

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

    The neuron is picking out identifier tokens—especially variable, constant, and function names—in code.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     and
    -1.88
     خمس
    -1.87
     zoude
    -1.85
    糴
    -1.81
     dieß
    -1.81
    殽
    -1.80
     BETTER
    -1.73
     увидел
    -1.69
    曏
    -1.69
    푆
    -1.68
    POSITIVE LOGITS
     $
    2.05
     of
    1.88
     time
    1.84
    )$
    1.75
     not
    1.73
     has
    1.71
    \
    1.66
     there
    1.64
     иногда
    1.60
     on
    1.59
    Activations Density 0.033%

    No Known Activations