© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 8897
    Prev
    Next
    INDEX
    Explanations

    Mets stadium fear GRID ways onomatopoeia

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

    This neuron fires strongly on the very first real word of a new section or paragraph—i.e. tokens that appear at the start of a fresh block of text.

    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
     interess
    0.75
    發生
    0.74
    ادث
    0.71
    ਸ
    0.69
    人は
    0.69
    ’
    0.68
     geomet
    0.68
     disse
    0.68
    น
    0.67
    vur
    0.66
    POSITIVE LOGITS
    <0x1C>
    0.80
    еты
    0.80
    ες
    0.77
     போன்றவை
    0.77
     Восто
    0.75
    которы
    0.75
    
    0.74
     Elkus
    0.73
    paper
    0.73
    adecimal
    0.73
    Activations Density 0.000%

    No Known Activations