© 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. 14042
    Prev
    Next
    INDEX
    Explanations

    all four, all beneficiaries

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

    The neuron fires on isolated, unusually high-valued single tokens that appear to be document or PDF headings and section titles (e.g. “Advances,” “Gravitation,” “Day,” “Wasser”), i.e. standalone capitalized words in metadata.

    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
    棍
    0.86
    anın
    0.81
    ९
    0.77
    后
    0.76
     secondaire
    0.75
     artesan
    0.75
    ۵
    0.75
     infest
    0.75
     inmedi
    0.74
    礁
    0.74
    POSITIVE LOGITS
    ри
    0.79
    ptime
    0.73
    сно
    0.70
     এছাড়া
    0.70
    ressed
    0.68
    ES
    0.66
     پ
    0.66
     shipped
    0.65
    rences
    0.64
     the
    0.63
    Activations Density 0.000%

    No Known Activations