© 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. 1840
    Prev
    Next
    INDEX
    Explanations

    levels, scales, and costs

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

    The neuron strongly activates on large numeric tokens (multi-digit numbers or hyphenated numeric ranges), effectively spotting high-magnitude figures.

    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
    cka
    0.57
    部に
    0.57
    стта
    0.56
    奴
    0.56
    ものが
    0.55
    絶対に
    0.55
    眼睛
    0.55
    敌人
    0.55
    ellipsis
    0.55
    の上に
    0.55
    POSITIVE LOGITS
    ‐
    0.86
     throughput
    0.77
    -/
    0.71
     precision
    0.71
     FIGURE
    0.67
     स्तरीय
    0.66
     termed
    0.66
    -$
    0.64
     programmatic
    0.63
     GAAP
    0.63
    Activations Density 0.469%

    No Known Activations