© 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. 59125
    Prev
    Next
    INDEX
    Explanations

    were the other/defending/most/two/main

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

    This neuron selectively activates on very common English function words (especially the definite article “the” and other high-frequency words like “were,” “only,” or “most”).

    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
     a
    -1.52
    Ꮨ
    -1.44
     pemasaran
    -1.39
    鄚
    -1.35
     oignon
    -1.33
    teilung
    -1.26
     eenen
    -1.23
    Kaip
    -1.23
    äksi
    -1.23
     an
    -1.23
    POSITIVE LOGITS
     ragazza
    1.41
     наших
    1.41
     extremamente
    1.39
     thing
    1.35
     fantástica
    1.33
     aislada
    1.29
     apresentar
    1.28
     deiner
    1.28
     see
    1.27
     was
    1.25
    Activations Density 0.049%

    No Known Activations