© 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. 10-GEMMASCOPE-RES-131K
    4. 80777
    Prev
    Next
    INDEX
    Explanations

    not working

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

    The neuron strongly responds to occurrences of the auxiliary “to be” (e.g. are, were, still be-type contexts) and associated passive constructions (often with negation like “not”).

    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
    ophageal
    -2.39
    های
    -2.19
    縐
    -2.19
    atellite
    -2.19
    ynthetic
    -2.14
     if
    -1.99
    hooter
    -1.95
    ynthesis
    -1.93
     dengan
    -1.93
     tica
    -1.90
    POSITIVE LOGITS
     to
    2.50
     neuesten
    2.17
     are
    2.17
     erwähnten
    2.17
     schönes
    2.13
     Pläne
    2.09
    嘡
    2.06
     trám
    2.05
    colgante
    2.05
    臵
    2.02
    Activations Density 0.392%

    No Known Activations