© 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. 101593
    Prev
    Next
    INDEX
    Explanations

    proper nouns and French words

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

    This neuron detects elided apostrophes in French tokens (the ’ character marking contractions like d’ or l’).

    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
     Numerous
    -3.08
    .”
    -3.02
    There
    -2.92
    It
    -2.84
     Shortly
    -2.83
     Apparently
    -2.78
     Ironically
    -2.78
     It
    -2.75
     nine
    -2.69
     Several
    -2.69
    POSITIVE LOGITS
    喾
    2.94
    ⎔
    2.94
     dezelve
    2.81
    𞤢
    2.73
    缯
    2.69
    錞
    2.67
    蹀
    2.61
     kommer
    2.61
     risiko
    2.59
     kuto
    2.58
    Activations Density 0.003%

    No Known Activations