© 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. 13478
    Prev
    Next
    INDEX
    Explanations

    areas, configuration, location

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

    The neuron strongly activates on content-rich or “information” tokens—i.e. meaningful nouns, adjectives, and other semantically important words rather than function words.

    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
     sekunder
    -1.03
    biete
    -0.95
    ieni
    -0.94
     farine
    -0.93
     magnum
    -0.91
     diep
    -0.90
     historien
    -0.89
     klink
    -0.88
     UNIVERS
    -0.88
     franc
    -0.86
    POSITIVE LOGITS
     proposed
    1.16
     new
    1.13
     location
    1.11
     locations
    1.08
     where
    0.95
     how
    0.93
     areas
    0.92
     potential
    0.86
     tłuma
    0.86
     different
    0.85
    Activations Density 0.079%

    No Known Activations