© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 64754
    Prev
    Next
    INDEX
    Explanations

    special sequences before punctuation

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

    The neuron selectively fires on the key content words that appear at the start of major sentences or paragraph openings—that is, the topic-setting nouns and verbs that kick off a new statement.

    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
     our
    -1.94
     we
    -1.80
     will
    -1.59
     would
    -1.58
     this
    -1.47
    We
    -1.45
     \&
    -1.45
     &
    -1.44
     was
    -1.43
    Our
    -1.35
    POSITIVE LOGITS
    kapital
    1.59
    jika
    1.43
    wendi
    1.33
    👥
    1.31
    ША
    1.31
    ignée
    1.31
    loggen
    1.30
    dokument
    1.30
     minimale
    1.30
    testen
    1.28
    Activations Density 0.002%

    No Known Activations