© 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. 22-GEMMASCOPE-RES-131K
    7. 39343
    Prev
    Next
    INDEX
    Explanations

    states or methods of completion

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

    The neuron activates strongly on past-participle verbs (especially in passive-voice constructions).

    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
     certamente
    -1.36
     vielä
    -1.35
    izzati
    -1.23
     łatwo
    -1.22
     facilmente
    -1.21
     participer
    -1.20
     vendeur
    -1.20
     erreich
    -1.20
     łat
    -1.20
     garantir
    -1.20
    POSITIVE LOGITS
     with
    3.78
     in
    2.58
     on
    2.52
     without
    2.06
     by
    1.90
     from
    1.89
     using
    1.80
     as
    1.80
     through
    1.57
     at
    1.54
    Activations Density 0.120%

    No Known Activations