© 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. 10-GEMMASCOPE-RES-131K
    7. 15846
    Prev
    Next
    INDEX
    Explanations

    technical terms and code

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

    The neuron fires on uncommon, content‐bearing tokens—proper names, technical terms, or rare/domain‐specific words—rather than on common function words.

    oai_token-act-pair · o4-mini
    Triggered 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
    Negative Logits
    <bos>
    -3.23
    圌
    -2.53
    あります
    -2.45
     (‘
    -2.42
    ’
    -2.38
    鶘
    -2.38
    providedIn
    -2.30
    kým
    -2.27
    却没有
    -2.25
    $
    -2.11
    POSITIVE LOGITS
    ي
    3.19
     девочка
    2.67
    anjutan
    2.61
    其妙
    2.56
    那麼
    2.47
    ...?
    2.47
    i
    2.47
    乆
    2.47
     faceva
    2.44
    er
    2.41
    Activations Density 0.003%

    No Known Activations