© 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. 3782
    Prev
    Next
    INDEX
    Explanations

    code definition or assignment

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

    The neuron fires most strongly on words occurring directly before a comma (i.e. clause-ending tokens).

    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
     ODDS
    0.92
     Masaki
    0.91
     T
    0.87
     R
    0.87
     H
    0.80
     Nonetheless
    0.79
    ﾏ
    0.79
     Kry
    0.79
     LE
    0.77
     AMS
    0.76
    POSITIVE LOGITS
    ز
    0.90
    ات
    0.83
    ن
    0.82
    bukti
    0.80
    imą
    0.79
    ა
    0.77
     زمانی
    0.75
    er
    0.74
    ске
    0.74
    ش
    0.71
    Activations Density 0.000%

    No Known Activations