© 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. 22-GEMMASCOPE-RES-131K
    7. 129081
    Prev
    Next
    INDEX
    Explanations

    The neuron consistently lights up on capitalized proper‐name tokens and abbreviations—e.g. institutions, personal names, place names, and similar named‐entity markers.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    St. followed by punctuation

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     thing
    -1.02
     maravilloso
    -1.00
    history
    -0.98
     antichi
    -0.98
     something
    -0.95
     luoghi
    -0.94
    ξ
    -0.93
    ョ
    -0.93
     everything
    -0.92
     réussi
    -0.92
    POSITIVE LOGITS
    visiae
    1.04
     forskning
    0.99
     research
    0.99
    żd
    0.98
    CTO
    0.98
    Läh
    0.98
    ־ה
    0.94
     البحث
    0.93
    Quantification
    0.93
    FONTE
    0.91
    Activations Density 0.027%

    No Known Activations