© 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. 3258
    Prev
    Next
    INDEX
    Explanations

    languages and locations

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

    The neuron consistently lights up on foreign‐language tokens or snippets (e.g. Italian, French, Czech, etc.), signaling detection of non-English text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    the word "of" when it appears in formal or official contexts, particularly in titles, organizational names, and governmental language.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    apostrophe-based clitics/elisions and similar punctuation patterns, especially in Romance-language text and time abbreviations.

    oai_token-act-pair · gpt-5Triggered 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
     potency
    0.58
    被称为
    0.57
     musí
    0.56
     endot
    0.56
     pesky
    0.56
     अधिकांश
    0.54
     most
    0.54
     kullanıl
    0.54
     används
    0.54
     toughest
    0.53
    POSITIVE LOGITS
     ২৮
    0.60
    ՝
    0.60
     dniu
    0.59
     tekint
    0.59
     solidar
    0.58
    Ду
    0.58
     Commune
    0.58
    舉行
    0.58
    popupButton
    0.57
    Lieutenant
    0.56
    Activations Density 0.024%

    No Known Activations