© 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. 68324
    Prev
    Next
    INDEX
    Explanations

    code configuration elements

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

    The neuron strongly activates on numeric literals—digits, sequences of digits (e.g. version numbers, dates, IDs), and associated punctuation—thus detecting number tokens.

    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
     still
    -1.22
    ‼
    -1.02
    XmlAccessorType
    -0.97
     なし
    -0.93
    龛
    -0.91
    燬
    -0.91
    🎬
    -0.90
     куртка
    -0.90
    <_>
    -0.90
     Despite
    -0.89
    POSITIVE LOGITS
     trastornos
    1.06
     desse
    1.02
     famoso
    1.00
    僰
    0.98
     diário
    0.97
     conquering
    0.97
    ふん
    0.96
     söyl
    0.96
    wór
    0.93
     bianche
    0.93
    Activations Density 0.023%

    No Known Activations