© 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. 35684
    Prev
    Next
    INDEX
    Explanations

    sentence beginnings and contractions

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

    The neuron strongly activates on numeric tokens and punctuation in numeric citations—i.e. years, volume/page numbers, and other digit sequences.

    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
     tendencias
    -0.83
    匕
    -0.81
     Tuesday
    -0.79
     прокура
    -0.79
    뚜
    -0.78
     Monday
    -0.77
    衍
    -0.77
     without
    -0.76
    ">
    -0.75
     obreros
    -0.75
    POSITIVE LOGITS
     getSession
    0.94
     applau
    0.89
     maya
    0.88
     dataSet
    0.87
     Penalty
    0.83
     favori
    0.82
     カジュアル
    0.81
     minif
    0.80
     vega
    0.79
     giù
    0.79
    Activations Density 0.018%

    No Known Activations