© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 6994
    Prev
    Next
    INDEX
    Explanations

    This neuron detects common sentence-initial function words or discourse markers (short closed-class words like “But,” “I,” “It,” “They,” “By,” etc.) that signal the start of a new clause or thought.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    therefore
    0.95
     dunque
    0.89
     therefore
    0.82
    Therefore
    0.79
    ename
    0.77
    จึง
    0.77
    ỗ
    0.74
     pertanto
    0.73
     infatti
    0.72
    ā
    0.72
    POSITIVE LOGITS
     نہیں
    0.93
     نہیں۔
    0.92
    并不
    0.89
     नहीं
    0.85
    并非
    0.84
    也不能
    0.84
     beware
    0.84
     ਨਹੀਂ
    0.84
     nejsou
    0.84
    只限
    0.83
    Activations Density 0.589%

    No Known Activations