© 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
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 102793
    Prev
    Next
    INDEX
    Explanations

    code blocks and declarations

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

    The neuron fires most strongly on structural code‐style tokens—things like braces (“{”), list dashes (“–”), uppercase identifiers (e.g. BYTE, Reserved, Init, Hint), and other punctuation used to mark out code blocks or config entries.

    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
    >"
    -0.98
    \""
    -0.85
     AppBundle
    -0.81
    claves
    -0.79
    Jumbo
    -0.77
     sprend
    -0.77
     cajas
    -0.77
    MenuItems
    -0.75
    œuf
    -0.74
    نجی
    -0.73
    POSITIVE LOGITS
    IZEN
    0.90
    urusan
    0.76
    chainId
    0.73
    oude
    0.72
     تعداد
    0.71
     bekommen
    0.68
     fung
    0.68
    irk
    0.67
     gebruik
    0.67
     nächsten
    0.67
    Activations Density 0.002%

    No Known Activations