© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 28207
    Prev
    Next
    INDEX
    Explanations

    authors and albums

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

    This neuron flags numeric quantifiers—particularly counts and figures (e.g., numbers of books, years, editions) in biographical or bibliographical contexts.

    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
     because
    -1.31
     to
    -1.24
     it
    -1.23
     newly
    -1.23
     new
    -1.21
     then
    -1.17
     their
    -1.15
     they
    -1.13
     provide
    -1.12
     just
    -1.09
    POSITIVE LOGITS
    %#
    1.27
     douane
    1.27
    himself
    1.26
     Sempre
    1.23
     négy
    1.20
    entitled
    1.20
     gouverneur
    1.20
    successfully
    1.20
    蚰
    1.19
    𓆏
    1.17
    Activations Density 0.047%

    No Known Activations