© 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. 34-GEMMASCOPE-RES-131K
    7. 2392
    Prev
    Next
    INDEX
    Explanations

    speaking, talking, saying

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

    The neuron strongly activates on word pieces that mean “speak,” “say,” “report,” or “speech” (e.g. English speak/speech, Slavic govor, Japanese いう, Lithuanian pranešėja, French rapporteur, etc.) across different languages.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     حصل
    -0.90
    鞏
    -0.80
     preguntas
    -0.80
     pregunta
    -0.78
    uckleberry
    -0.76
    ಂದು
    -0.75
     plainte
    -0.75
     darling
    -0.73
     question
    -0.73
    relse
    -0.73
    POSITIVE LOGITS
    ची
    0.96
    🗣
    0.90
    ativeness
    0.89
     comerciales
    0.87
    ShowDialog
    0.87
     Hype
    0.87
    mouth
    0.84
    battery
    0.83
    清晨
    0.82
    boxes
    0.82
    Activations Density 0.006%

    No Known Activations