© 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. 34-GEMMASCOPE-RES-131K
    7. 118095
    Prev
    Next
    INDEX
    Explanations

    code blocks with angle brackets

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

    The neuron fires on structural punctuation in code or markup (e.g. braces, semicolons, angle brackets, parentheses, underscores), i.e. non-letter symbols that form programming or XML 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
     farine
    -0.97
    あなたが
    -0.91
     puissiez
    -0.88
     Docteur
    -0.86
    worpen
    -0.86
    咔
    -0.85
    ：　
    -0.84
     profuse
    -0.82
     Prist
    -0.82
    днев
    -0.81
    POSITIVE LOGITS
     TResult
    0.87
     rounds
    0.86
    woocommerce
    0.81
     synthesis
    0.79
     ____
    0.79
    forall
    0.77
    elasticsearch
    0.77
     imagin
    0.76
     composition
    0.75
     representative
    0.75
    Activations Density 0.020%

    No Known Activations