© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 79582
    Prev
    Next
    INDEX
    Explanations

    noun−noun phrases

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

    The neuron strongly activates on repeated two-word compound noun phrases (e.g. “assisted living,” “hallway table,” “custom engagement ring,” etc.).

    oai_token-act-pair · o4-miniTriggered 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
     Bewußt
    -1.54
     quí
    -1.46
     lã
    -1.41
     cerâmica
    -1.40
     soldadura
    -1.38
     desactivar
    -1.36
     jsem
    -1.36
     repuesto
    -1.31
     trám
    -1.31
     aguj
    -1.30
    POSITIVE LOGITS
     for
    1.47
     in
    1.27
    r
    1.21
     during
    1.17
     augusti
    1.14
     of
    1.13
     your
    1.09
     Kommune
    1.05
     maroc
    1.03
     iyong
    1.02
    Activations Density 0.054%

    No Known Activations