© 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. 31-GEMMASCOPE-2-RES-16K
    7. 6630
    Prev
    Next
    INDEX
    Explanations

    The neuron selectively activates on numeric tokens that include digits and decimal points (e.g. percentages, millions, billions, etc.).

    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_31_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
    E
    0.62
    Z
    0.62
     Btu
    0.61
    V
    0.56
    B
    0.55
     mL
    0.54
    প্র
    0.54
     দশমিক
    0.53
     Drying
    0.53
    c
    0.52
    POSITIVE LOGITS
    ジネス
    0.63
    infodisc
    0.60
     religiosos
    0.58
    ᓵ
    0.57
     difficiles
    0.56
    kfollowers
    0.55
    麼
    0.55
     similaires
    0.55
    审查
    0.55
     jaanu
    0.55
    Activations Density 0.100%

    No Known Activations