© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 1626
    Prev
    Next
    INDEX
    Explanations

    contrast or exception

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

    The neuron fires on dense, formal scientific or technical prose laden with multi‐syllabic jargon, numeric values, and research-style discourse.

    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
    zięki
    0.73
    なら
    0.67
     übrigens
    0.66
    なくなった
    0.63
     EVERY
    0.62
    これで
    0.62
     beautifully
    0.62
     sayesinde
    0.60
     superbly
    0.60
    就可以了
    0.59
    POSITIVE LOGITS
     difficulties
    1.84
     lacks
    1.69
    缺乏
    1.66
    很难
    1.64
    难以
    1.61
     lacked
    1.60
     difficoltà
    1.57
     difficulty
    1.55
    difficult
    1.55
     lack
    1.55
    Activations Density 0.996%

    No Known Activations