© 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. 22-GEMMASCOPE-RES-131K
    7. 128699
    Prev
    Next
    INDEX
    Explanations

    The neuron is highly sensitive to numerals and numeric references (especially years, dates, and measurements) in encyclopedic/descriptive text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    first described by

    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
    Not in Any Lists

    No Comments

    Negative Logits
    PROLOG
    -1.02
    www
    -1.02
     būs
    -0.98
    鄙
    -0.96
     Verhältnis
    -0.95
    aien
    -0.93
     izvē
    -0.93
    wiście
    -0.93
    iteracy
    -0.92
     celý
    -0.92
    POSITIVE LOGITS
     by
    1.74
     after
    1.38
     magasiner
    1.16
     as
    1.12
     when
    1.08
    καν
    1.06
     in
    1.02
    вець
    1.02
    vielen
    1.00
     referirse
    0.99
    Activations Density 0.104%

    No Known Activations