© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 1410
    Prev
    Next
    INDEX
    Explanations

    Arabic prefixes and subsequent words

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

    The neuron activates strongly on non-English tokens—i.e. words in other scripts (Arabic, Hebrew) or with non-ASCII/diacritic characters—so it’s detecting foreign-language or accented text.

    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
    에
    0.71
    0.64
    我
    0.61
    看来
    0.61
    И
    0.59
     mỗi
    0.58
    는
    0.57
    岖
    0.57
    내
    0.56
     Marlborough
    0.55
    POSITIVE LOGITS
    ور
    0.77
    ست
    0.76
    v
    0.72
     اغ
    0.71
    عمل
    0.71
    يك
    0.70
    راث
    0.70
    ूथ
    0.69
    فاع
    0.68
    سات
    0.68
    Activations Density 0.000%

    No Known Activations