© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 129064
    Prev
    Next
    INDEX
    Explanations

    The neuron detects literal tokens in code—numeric constants and punctuation‐only or symbol‐only string literals.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    checking variable conditions

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     スリーブ
    -1.11
    瘆
    -0.93
    については
    -0.93
     valentino
    -0.93
    鮰
    -0.90
    Masa
    -0.89
     lâng
    -0.88
    驮
    -0.88
     extravag
    -0.88
    はじめて
    -0.88
    POSITIVE LOGITS
     or
    1.23
     Почему
    0.99
     there
    0.94
     in
    0.93
     nearing
    0.90
     as
    0.90
     פאר
    0.90
     when
    0.89
     null
    0.88
     część
    0.88
    Activations Density 0.005%

    No Known Activations