© 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. 105623
    Prev
    Next
    INDEX
    Explanations

    multi-script fragments

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

    The neuron activations all occur on subword fragments that serve as suffixes or inflectional endings (e.g. verb endings, clitic attachments, possessive ’s), so it’s detecting word‐ending pieces.

    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
     spez
    -1.49
     Macromolecules
    -1.37
     Privac
    -1.36
     mitsubishi
    -1.34
     haup
    -1.34
     LIBRA
    -1.34
     ?????
    -1.34
     singola
    -1.27
     Intere
    -1.27
     maer
    -1.27
    POSITIVE LOGITS
    <blockquote>
    1.59
     İş
    1.45
    It
    1.36
     to
    1.30
     the
    1.26
     системой
    1.24
    реб
    1.23
    matory
    1.23
     not
    1.20
    Lol
    1.20
    Activations Density 0.011%

    No Known Activations