© 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. 12608
    Prev
    Next
    INDEX
    Explanations

    device :0

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

    The neuron fires on numerical and code‐style tokens—especially digit sequences and numeric literals (indices, version numbers, addresses, etc.) in technical or programming contexts.

    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
    Buyer
    0.57
     сообщи
    0.55
     휴
    0.55
     Hooper
    0.55
    orro
    0.54
    candy
    0.53
    posted
    0.52
    Posting
    0.52
    Candy
    0.52
     이러
    0.52
    POSITIVE LOGITS
    ０
    1.36
     zeroth
    1.26
     ০
    1.23
    0
    1.22
    ᱐
    1.12
    ០
    1.09
    𝟎
    1.09
    ₀
    1.07
    ۰
    1.02
    ٠
    1.01
    Activations Density 0.032%

    No Known Activations