© 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. 10-GEMMASCOPE-RES-131K
    7. 100294
    Prev
    Next
    INDEX
    Explanations

    code delimiters

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

    text following a colon and space, typically in code comments, string literals, or label-value pairs.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron activates on numeric constants (integer literals) in code or 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_10/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
    .
    -3.38
    茕
    -2.67
    a
    -2.66
    r
    -2.64
    ve
    -2.59
    they
    -2.55
     parachoque
    -2.52
    l
    -2.47
    푣
    -2.39
    for
    -2.38
    POSITIVE LOGITS
    er
    2.91
     Hinblick
    2.86
     Traditionally
    2.64
     tornou
    2.58
     Вода
    2.52
     Indeed
    2.47
     finds
    2.47
     Therefore
    2.44
     scrutin
    2.44
     Notably
    2.42
    Activations Density 0.002%

    No Known Activations