© 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. 10-GEMMASCOPE-RES-131K
    7. 73477
    Prev
    Next
    INDEX
    Explanations

    script file

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

    The neuron fires on domain‐specific jargon and proper‐name tokens—things like technology names, file or format names, component labels, chemical compound names, abbreviations, version numbers, and similar technical terms.

    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
     are
    -1.80
     jsme
    -1.55
     were
    -1.55
     how
    -1.52
     были
    -1.44
     których
    -1.41
     who
    -1.41
    There
    -1.41
     ćwic
    -1.40
    ਫ਼
    -1.40
    POSITIVE LOGITS
     ۱۱
    1.47
    hradun
    1.46
    this
    1.46
     to
    1.43
    ronomie
    1.37
     ۱۵
    1.37
     ۱۲
    1.35
     beton
    1.34
    对你
    1.31
     bula
    1.29
    Activations Density 0.251%

    No Known Activations