© 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-3-27B-IT
    5. 11-GEMMASCOPE-2-TRANSCODER-262K
    7. 90084
    Prev
    Next
    INDEX
    Explanations

    * "world-renowned" followed by a profession/field* renowned expert/professional* detects "world-renowned" followed by noun* fame and expertise descriptorsThe explanation should be *just* the pattern itself, 3-20 words."renowned" appears consistently in MAX_ACTIVATING_TOKENS."world-renowned" appears in TOP_ACTIVATING_TEXTS.The tokens after are diverse fields/professions.The simplest explanation capturing this is how "renowned" is used, often qualified by "world."Let's consider the prompt's examples:- "unique words" - too general.- "foreign words" - too general.- "words starting with a sequence" is a *type* of pattern.The pattern here is the word "renowned" itself, often qualified by "world," and followed by specific types of nouns (professions, places, experts).How to make it specific and concise?"prestigious achievements" - maybe too interpretive."highly respected fields" - good, but maybe less

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_11_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    targetReference
    0.77
    ెంట్‌
    0.76
    on
    0.73
     nCenters
    0.73
    eryka
    0.73
    كو
    0.70
    tól
    0.70
     ጥቅ
    0.68
    مة
    0.67
    ﺔ
    0.67
    POSITIVE LOGITS
    ↵
    0.84
     b
    0.71
    0.70
     V
    0.64
    s
    0.63
     Very
    0.62
    b
    0.61
     [
    0.60
    ↵↵
    0.60
     (
    0.59
    Activations Density 0.001%

    No Known Activations