© 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. 65156
    Prev
    Next
    INDEX
    Explanations

    numbers and timestamps

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

    The neuron is triggered by numeric tokens—particularly multi-digit sequences such as timestamps, IDs, version numbers, and similar long numbers.

    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
    為に
    -1.48
    چرخ
    -1.18
     udaler
    -1.16
    ようです
    -1.11
    toronto
    -1.05
    垕
    -1.05
     stadt
    -1.04
    chauff
    -1.04
    </h5>
    -1.03
     it
    -1.03
    POSITIVE LOGITS
    <bos>
    1.73
    というわけで
    1.49
    結局
    1.43
    個人的には
    1.40
    ionales
    1.36
    essment
    1.34
    もしか
    1.34
    という感じ
    1.33
    ymmetry
    1.33
    それにしても
    1.30
    Activations Density 0.067%

    No Known Activations