© 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. 128606
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on the key “attribute” words you see in factual or infobox‐style lines—things like measurement terms (diameter, kilometers), section headings or property names (Orbit, classification, games, County), and the copular “is/has” that introduce them—i.e. it spotlights the labels that carry concrete data in encyclopedia‐style entries.

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

    historical locations and people

    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
     one
    -1.41
     three
    -1.41
     two
    -1.39
     our
    -1.37
     four
    -1.34
     many
    -1.29
     this
    -1.27
     five
    -1.21
     all
    -1.20
     specific
    -1.17
    POSITIVE LOGITS
     exemplos
    1.40
    釕
    1.27
     patrones
    1.24
     padrões
    1.20
     vetores
    1.19
     facultad
    1.17
     ocultar
    1.16
     mensajes
    1.15
    ὺ
    1.14
     reflexiones
    1.13
    Activations Density 0.001%

    No Known Activations