© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 10383
    Prev
    Next
    INDEX
    Explanations

    clarifications or alternatives

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

    The neuron strongly activates on conversational function words and discourse markers (e.g. “well,” “actually,” “rather,” pronouns like “you”), i.e. filler and personal-address 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
     (
    0.59
    fig
    0.55
     fig
    0.54
     other
    0.54
     typographic
    0.54
     fluorescence
    0.52
    autres
    0.52
    либо
    0.51
    <sup>
    0.50
     vigorous
    0.50
    POSITIVE LOGITS
     Actually
    0.84
    actually
    0.81
     eigentlich
    0.78
     एक्चुअली
    0.71
     egent
    0.70
     sebenarnya
    0.69
    Actually
    0.69
     البته
    0.69
     einzigen
    0.69
     actually
    0.68
    Activations Density 0.132%

    No Known Activations