© 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. 13983
    Prev
    Next
    INDEX
    Explanations

    potentially coercive, influence writing, interpreted as

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

    The neuron fires on structural or metadata tokens—things like section headings, numbered list titles, citation entries, dates, and other document‐formatting markers.

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

    very high numerical values, especially large numbers in the hundreds.

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

    document headings, headlines, and bibliographic/reference metadata within nonfiction text.

    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
     lysate
    0.93
     Sheng
    0.93
     rinsing
    0.93
     rationalize
    0.92
     detergents
    0.90
     camphor
    0.90
    тной
    0.89
     Río
    0.89
     enteric
    0.88
     carnivorous
    0.87
    POSITIVE LOGITS
    s
    1.05
    i
    0.83
    ui
    0.82
    👇
    0.81
    ob
    0.79
    un
    0.78
    no
    0.77
    에
    0.76
    ug
    0.76
    na
    0.76
    Activations Density 0.001%

    No Known Activations