© 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. 1845
    Prev
    Next
    INDEX
    Explanations

    literally a

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

    The neuron is primarily firing on content words that describe dynamic actions or vivid sensory events (especially verbs in –ing or past‐participle form and other high-salience nouns) rather than on function words.

    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
     IUnary
    0.80
     ausreichend
    0.77
     QTL
    0.77
     INumber
    0.75
     EPFO
    0.74
     Gautam
    0.74
     hashCode
    0.70
     `>=
    0.68
     IEquatable
    0.68
     sufficient
    0.67
    POSITIVE LOGITS
    表演
    0.82
    fake
    0.80
     antics
    0.77
    いろんな
    0.75
    战士
    0.75
     acrob
    0.73
    唱歌
    0.73
    さまざまな
    0.72
    amazing
    0.71
    incredible
    0.70
    Activations Density 0.003%

    No Known Activations