© 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. 8076
    Prev
    Next
    INDEX
    Explanations

    modifiers before words

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

    The neuron activates on terms that label hierarchical subdivisions or categories (e.g. “sub-”, “major/minor,” “secondary,” “aggregate,” “categories”).

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

    the word "sub" when it appears as a prefix or in hierarchical/categorical contexts (like "sub-categories", "sub-regional", "subtests").

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

    mentions of hierarchical structure—subdivisions, levels (major/minor, upper/lower, primary/secondary), and aggregate groupings within a system.

    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
     könnte
    0.71
    uppe
    0.68
    स्तान
    0.65
     away
    0.64
     konnte
    0.64
    ador
    0.63
    people
    0.63
    no
    0.61
     bisa
    0.61
    sı
    0.61
    POSITIVE LOGITS
     subdivided
    0.76
     වශයෙන්
    0.70
     spezif
    0.68
    分为
    0.65
    する場合は
    0.64
     specialized
    0.63
     Specify
    0.63
     subgroups
    0.62
     अर्थात्
    0.61
     これらの
    0.61
    Activations Density 0.026%

    No Known Activations