© 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. 34186
    Prev
    Next
    INDEX
    Explanations

    code suffixes

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

    The neuron responds to code and technical identifiers—variable/class names, domain‐specific terms, and numeric literals—i.e. tokens that look like programming identifiers or technical jargon.

    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
     now
    -1.30
     Z
    -1.06
     before
    -1.02
     neighboring
    -0.96
     M
    -0.94
     probably
    -0.93
     J
    -0.93
     S
    -0.90
     mainly
    -0.90
     William
    -0.88
    POSITIVE LOGITS
     エントランス
    1.07
     motiva
    1.07
     primario
    1.06
    ﮨ
    1.06
     indiví
    1.03
     kompeti
    1.03
     livros
    1.02
     presentada
    1.01
     unterkunft
    1.00
     movimenta
    1.00
    Activations Density 0.140%

    No Known Activations