© 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. 15437
    Prev
    Next
    INDEX
    Explanations

    titles and descriptions

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

    The neuron fires strongly on named titles or headings—e.g. program or project names and document or session titles (often in title case or in quotes).

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

    Let me analyze the activation patterns for Neuron 4. Looking at the high activation values (>200): First document: Very high activations on words like "supports" (363.7), "Expansion" (383.8), "of" (421.9),

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

    titles and formal content

    np_max-act-logits · claude-4-5-sonnetTriggered by @jrosseruk
    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
     billions
    0.75
    Also
    0.75
    groups
    0.73
    ardır
    0.71
    charts
    0.71
    illions
    0.70
     constituencies
    0.70
     pundits
    0.69
    ratings
    0.69
    liners
    0.68
    POSITIVE LOGITS
     Urinary
    0.86
     Fertility
    0.85
     Floral
    0.84
     Dissertation
    0.84
     Mitochondrial
    0.84
     Fire
    0.83
     Insect
    0.82
     Oste
    0.81
     Rheumat
    0.81
     Tuberculosis
    0.81
    Activations Density 2.106%

    No Known Activations