© 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. 13779
    Prev
    Next
    INDEX
    Explanations

    agents, briefs, ASOS

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

    The neuron strongly activates on small functional tokens in progressive or future‐tense constructions—auxiliary verbs like “will,” forms of “be,” the gerund “helping,” and the adverb “up.”

    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
    但我
    0.85
    但他
    0.68
     drei
    0.67
     AcOH
    0.66
    以外の
    0.66
     لكن
    0.64
     OnePlus
    0.62
    🖋
    0.62
     THREE
    0.61
     tiga
    0.61
    POSITIVE LOGITS
    非常的
    0.70
    which
    0.67
    that
    0.64
    çok
    0.64
     szczegól
    0.63
    trong
    0.63
    也非常
    0.63
     学生
    0.63
    ennen
    0.63
    ckså
    0.63
    Activations Density 0.221%

    No Known Activations