© 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. 100153
    Prev
    Next
    INDEX
    Explanations

    multi-script and specialized terms like gunshot

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

    This neuron responds to sibilant “sh”-type letters with diacritics or in non-Latin scripts (e.g. Latin “ş/Ş,” Cyrillic “ш,” Arabic “ش”).

    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
    樘
    -2.19
     Acuerdo
    -2.16
    髄
    -2.11
     criaturas
    -2.06
    ۉ
    -2.03
     mistaken
    -2.03
     urge
    -2.03
     roundabout
    -2.02
     universitarios
    -2.00
    很
    -2.00
    POSITIVE LOGITS
     ش
    2.33
    z
    2.25
    有两
    2.22
     Grüßen
    2.17
    ).
    2.11
     és
    2.09
    x
    2.09
    莇
    2.08
    那我
    2.08
    beten
    2.08
    Activations Density 0.001%

    No Known Activations