© 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. 10-GEMMASCOPE-RES-131K
    7. 105313
    Prev
    Next
    INDEX
    Explanations

    according CD, for example

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

    The neuron fires on formatting and structural tokens in technical or academic text—especially standalone parentheses, hyphens, and numeric strings.

    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
     économies
    -1.23
    搴
    -1.21
    “:
    -1.20
     diki
    -1.20
     SOMETHING
    -1.18
    延长
    -1.17
    Cartney
    -1.16
     glut
    -1.16
    "../
    -1.16
    leuth
    -1.16
    POSITIVE LOGITS
    Exeunt
    1.57
    だよ
    1.34
     is
    1.31
     χρησιμοποι
    1.28
     Еще
    1.26
    المل
    1.23
     kolejny
    1.21
    <?
    1.17
    内核
    1.17
     живота
    1.16
    Activations Density 0.318%

    No Known Activations