© 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. 58880
    Prev
    Next
    INDEX
    Explanations

    linguistics or proper nouns

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

    The neuron is picking out capitalized tokens—proper names, acronyms, and other words that begin with an uppercase letter.

    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
    s
    -3.77
    螄
    -3.28
    er
    -2.63
     插图
    -2.47
     uruguay
    -2.47
     prenn
    -2.34
     itſelf
    -2.33
    this
    -2.31
     современного
    -2.30
    ，“
    -2.27
    POSITIVE LOGITS
    rierten
    2.64
     afirma
    2.63
     Commonly
    2.59
     najwięks
    2.56
     często
    2.53
     ocurrir
    2.50
     französische
    2.41
     characterize
    2.39
     critique
    2.36
     sigue
    2.34
    Activations Density 0.003%

    No Known Activations