© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 40919
    Prev
    Next
    INDEX
    Explanations

    /dev/

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

    The neuron strongly activates on numeric tokens (e.g. version numbers, device IDs, counts)—in other words, it detects number literals.

    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
    _
    -2.14
    >`;
    -1.74
    st
    -1.62
    [
    -1.53
     was
    -1.50
    final
    -1.49
    ba
    -1.48
    can
    -1.48
     culminating
    -1.45
    ro
    -1.45
    POSITIVE LOGITS
    豸
    1.71
    homonymie
    1.67
     which
    1.55
    Surprisingly
    1.55
    ющая
    1.55
    แนะนำ
    1.52
    Unlike
    1.48
     הם
    1.48
    Kiki
    1.47
    ógico
    1.45
    Activations Density 0.015%

    No Known Activations