© 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. 84536
    Prev
    Next
    INDEX
    Explanations

    quotation marks and titles

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

    The neuron fires on isolated three‐digit numeric tokens (e.g. “430,” “410,” “370”)—in other words, it detects page numbers or similar numeric identifiers in citation and reference sections.

    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
     if
    -1.55
    也會
    -1.47
     zelve
    -1.45
    来週
    -1.45
    明日は
    -1.44
    になってしまう
    -1.43
    alló
    -1.42
    たびに
    -1.42
     any
    -1.42
    昨日の
    -1.41
    POSITIVE LOGITS
    があり
    1.76
     (
    1.57
     the
    1.55
    '
    1.46
    an
    1.45
    у
    1.44
    ,
    1.41
     siècles
    1.41
    но
    1.40
    <>
    1.40
    Activations Density 0.013%

    No Known Activations