© 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. 22-GEMMASCOPE-RES-131K
    4. 68201
    Prev
    Next
    INDEX
    Explanations

    WebAcl or Str

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

    This neuron is a “code detector” – it fires on non-prose tokens typical of program source (curly braces, import/using/namespace/package keywords, all-caps identifiers, version numbers, 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
     their
    -1.66
     this
    -1.66
     people
    -1.64
     first
    -1.59
     our
    -1.49
     strong
    -1.38
     well
    -1.38
     more
    -1.38
     unbelievable
    -1.37
     incredible
    -1.36
    POSITIVE LOGITS
     ORO
    1.27
     ANTI
    1.20
     uly
    1.19
    糁
    1.19
    hese
    1.19
     BUILT
    1.16
    鮃
    1.16
     tronic
    1.15
     ſeveral
    1.15
     uſ
    1.14
    Activations Density 0.019%

    No Known Activations