© 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. 9743
    Prev
    Next
    INDEX
    Explanations

    license text

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

    The neuron fires on the first “content” word of a new question or directive (e.g. “How long,” “What’s,” “free,” “Eat That Frog,” “even,” “familiar,” etc.), i.e. it marks the entry point of questions and calls‐to‐action.

    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
    Regardless
    0.59
    動力
    0.59
    基金
    0.57
    月的
    0.57
    玦
    0.57
    ▢
    0.57
    WEBPACK
    0.55
     descob
    0.55
     கிறிஸ்துவின்
    0.55
    拥有
    0.55
    POSITIVE LOGITS
     knitted
    0.56
    рованное
    0.55
    ాయని
    0.55
     bei
    0.54
     budgets
    0.54
    рованные
    0.53
     blueberries
    0.52
     \%
    0.51
     YOUR
    0.50
    uhn
    0.49
    Activations Density 0.000%

    No Known Activations