© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 95664
    Prev
    Next
    INDEX
    Explanations

    place names in France and Russia

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

    This neuron detects tokens that are part of multi‐word administrative place names (especially hyphenated or accented commune/canton names).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     here
    -1.25
     will
    -1.17
     tarte
    -1.13
     białym
    -1.12
    ようになります
    -1.06
    пасибо
    -1.05
     begin
    -1.04
     vlastnosti
    -1.04
     vektör
    -1.03
     graden
    -1.03
    POSITIVE LOGITS
    Leider
    1.41
     Obwohl
    1.35
    Haben
    1.27
    hradun
    1.21
    まい
    1.20
     Aufgrund
    1.19
    chanical
    1.18
     行っ
    1.17
     recibiendo
    1.17
    vants
    1.16
    Activations Density 0.011%

    No Known Activations