© 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. 22-GEMMASCOPE-RES-131K
    7. 14733
    Prev
    Next
    INDEX
    Explanations

    surgical procedures

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

    The neuron identifies programming code syntax, particularly in brackets, parentheses, and technical formatting contexts.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    This neuron responds to tokens containing digits, i.e. numeric values (version numbers, percentages, counts, etc.).

    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
    Boas
    -0.99
    逻
    -0.92
    Acerca
    -0.91
    arkhand
    -0.89
    Dota
    -0.88
    Romance
    -0.84
     gruesa
    -0.84
    しと
    -0.84
    Herzlich
    -0.82
     pierwszej
    -0.82
    POSITIVE LOGITS
     polie
    1.08
    <b>
    0.97
     garan
    0.96
     plafon
    0.94
     fuga
    0.93
    чье
    0.91
     variante
    0.90
     AMOUNT
    0.90
     ordine
    0.89
     ティー
    0.89
    Activations Density 0.013%

    No Known Activations