© 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. 68683
    Prev
    Next
    INDEX
    Explanations

    UI elements and technical configurations

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

    The neuron is picking out longer, multi‐syllabic technical or domain‐specific content words (e.g. “expansion,” “panels,” “vertical,” “disengages,” 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
     durer
    -1.57
     célè
    -1.51
     choisir
    -1.48
     venant
    -1.46
     mépris
    -1.45
     régal
    -1.37
     THROUGH
    -1.36
    ჻
    -1.36
     légèrement
    -1.35
     fidé
    -1.35
    POSITIVE LOGITS
     or
    1.75
     with
    1.73
     this
    1.66
     on
    1.63
     can
    1.56
     has
    1.56
     as
    1.50
     serves
    1.47
     just
    1.47
     before
    1.46
    Activations Density 0.307%

    No Known Activations