© 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. 22-GEMMASCOPE-RES-131K
    7. 12875
    Prev
    Next
    INDEX
    Explanations

    programming code

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

    The neuron fires strongly on programming‐language punctuation and operator tokens (brackets, equals signs, function keywords, etc.), essentially detecting code syntax rather than natural‐language text.

    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
    硖
    -1.49
    を変更
    -1.46
    全体的に
    -1.46
    regler
    -1.39
    সূত্র
    -1.38
     Reised
    -1.38
    onekana
    -1.37
     tidligere
    -1.34
     previous
    -1.30
    bluza
    -1.28
    POSITIVE LOGITS
     how
    1.29
    ｗ
    1.18
     сочи
    1.17
     яйцо
    1.16
     Harus
    1.14
     from
    1.14
    Վ
    1.10
     unusually
    1.09
    εται
    1.09
     Especially
    1.09
    Activations Density 0.110%

    No Known Activations