© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 7236
    Prev
    Next
    INDEX
    Explanations

    AI ethics and safety boundaries

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

    This neuron fires most strongly on the distinctive, content-bearing nouns or proper names that conclude section headings or standalone lines (e.g. “Railway,” “expenses,” “Friday,” “agencies,” “skippers,” “Machinery”).

    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
    s
    0.82
    zione
    0.75
    t
    0.73
    Movie
    0.73
    '
    0.73
    น
    0.71
    tant
    0.71
    ت
    0.71
    nt
    0.70
    1
    0.68
    POSITIVE LOGITS
    쭌
    0.89
     cramping
    0.88
    뀨
    0.86
     хоро
    0.84
    𝚇
    0.82
     เออ
    0.82
     Мы
    0.81
     hãng
    0.81
     inade
    0.80
    슌
    0.79
    Activations Density 0.000%

    No Known Activations