© 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. 13134
    Prev
    Next
    INDEX
    Explanations

    multiplication before addition or subtraction

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

    The neuron strongly activates on salient content words—especially concrete nouns and acronyms—while ignoring most function words.

    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
    エット
    0.87
     או
    0.80
     digitar
    0.80
    Ւ
    0.80
     اقدام
    0.79
     ogran
    0.77
     たり
    0.77
     ovlád
    0.76
     impedir
    0.75
     Следу
    0.73
    POSITIVE LOGITS
    y
    1.02
    ли
    0.89
    t
    0.87
    pins
    0.86
    les
    0.84
    er
    0.82
    ுள்ளார்
    0.80
    es
    0.80
    dang
    0.80
    eight
    0.79
    Activations Density 0.000%

    No Known Activations