© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 32681
    Prev
    Next
    INDEX
    Explanations

    code blocks and parentheses

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

    This neuron detects code-structure punctuation and whitespace—things like parentheses, braces, semicolons, and indentation that define programming syntax.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     still
    -0.92
     first
    -0.91
     Loren
    -0.90
     as
    -0.88
     one
    -0.88
    集合
    -0.86
    ここ
    -0.85
    十四
    -0.84
     urged
    -0.84
    Ret
    -0.84
    POSITIVE LOGITS
    public
    1.30
     public
    0.99
    /></
    0.98
     IJ
    0.84
     tarafından
    0.84
    ilter
    0.83
     Mileage
    0.83
    ())){
    0.82
     LeBron
    0.81
     meubles
    0.81
    Activations Density 0.002%

    No Known Activations