© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 127888
    Prev
    Next
    INDEX
    Explanations

    The neuron is keyed to constant and literal tokens in code—things like numeric literals (e.g. 4, 200), enum or constant names (e.g. ELEMENT_NODE, EMAIL_ADDRESS), and the punctuation around them (parentheses, case labels, etc.).

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408

    ; bracket and function delimiters

    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
    Negative Logits
    Commune
    -0.90
    werve
    -0.90
     the
    -0.83
    pushButton
    -0.83
    Rt
    -0.79
     ert
    -0.79
     vasca
    -0.78
    stt
    -0.78
     mystère
    -0.77
     Knows
    -0.76
    POSITIVE LOGITS
     or
    1.08
     grijs
    1.02
    しかも
    1.02
    ându
    0.98
    其
    0.93
     its
    0.90
    ökning
    0.90
     blauw
    0.88
     sección
    0.86
    化粧
    0.86
    Activations Density 0.007%

    No Known Activations