© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 72445
    Prev
    Next
    INDEX
    Explanations

    descriptors followed by place types

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

    The neuron strongly responds to capitalized place names and venues marked by words like “National,” “State,” or “International” (e.g. parks and airports).

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

    the word "National" when it precedes "Park" or appears in official names of parks, airports, and institutions.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    capitalized institutional designators within proper names (e.g., modifiers in names of parks, airports, organizations, and facilities).

    oai_token-act-pair · gpt-5Triggered by @jyhe0408
    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
    !
    -1.45
     that
    -1.37
     are
    -1.35
     they
    -1.30
     also
    -1.23
     utara
    -1.21
    うれ
    -1.20
     halben
    -1.20
    喜んで
    -1.19
    ほら
    -1.18
    POSITIVE LOGITS
    <bos>
    1.34
    ارتمان
    1.16
    說道
    1.16
     manier
    1.16
     stolet
    1.15
    结语
    1.14
     những
    1.13
     contenant
    1.12
     illustratie
    1.12
    pography
    1.12
    Activations Density 0.067%

    No Known Activations