© 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-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 65228
    Prev
    Next
    INDEX
    Explanations

    adverbs of truthfulness and seriousness

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

    The neuron activates most strongly on discourse-marking adverbs (especially “honestly,” and to a lesser extent “seriously”) used to convey speaker emphasis or sincerity.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Langkah
    -1.84
     рассказать
    -1.66
    ’，
    -1.62
     niedersachsen
    -1.62
     Однако
    -1.62
    Также
    -1.60
     karena
    -1.59
    товых
    -1.59
    ће
    -1.51
    jaket
    -1.51
    POSITIVE LOGITS
    a
    1.76
    lossians
    1.76
    <bos>
    1.72
    ziehungs
    1.65
    henswürdigkeiten
    1.64
    ную
    1.62
    wości
    1.61
     which
    1.57
    ter
    1.56
    quelize
    1.56
    Activations Density 0.002%

    No Known Activations