© 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. 130666
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on open-class (“content”) words—nouns, verbs, adjectives, pronouns—rather than on function words like articles or conjunctions.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    German and English mixed language

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     and
    -1.70
     daardoor
    -1.09
     بسیار
    -1.04
     allzu
    -1.02
     vielmehr
    -1.02
     increases
    -1.00
     this
    -0.99
    Outras
    -0.98
    eabouts
    -0.98
     give
    -0.98
    POSITIVE LOGITS
     also
    2.23
     already
    1.42
     now
    1.35
     nowadays
    1.33
     indeed
    1.23
    also
    1.23
     moreover
    1.17
    も
    1.17
    もある
    1.16
    も多い
    1.13
    Activations Density 0.006%

    No Known Activations