© 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-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 93427
    Prev
    Next
    INDEX
    Explanations

    possibly, probably, perhaps

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

    This neuron is triggered by discourse-level markers and parenthetical punctuation—things like “probably,” “actually,” contrastive “not,” “ever,” and opening parentheses or slashes that flag hedges, emphatic asides, and other sentence-level qualifiers.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     entwick
    -1.06
    always
    -0.94
    EMPLOYMENT
    -0.92
    ığımız
    -0.90
     entwickelte
    -0.86
    みたい
    -0.85
     ответы
    -0.84
    tação
    -0.82
    Mbps
    -0.81
     obicei
    -0.81
    POSITIVE LOGITS
     maybe
    1.34
     possibly
    1.27
     probably
    1.17
     comparisons
    1.10
     considering
    1.09
     perhaps
    1.09
    Possibly
    1.05
    Probably
    0.95
    Considering
    0.95
     Possibly
    0.94
    Activations Density 0.029%

    No Known Activations