© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 129513
    Prev
    Next
    INDEX
    Explanations

    This neuron fires on structural document labels and headings, such as numbered sections (“One,” “Two,” “Part I/II”), experiments (“Experiment 1/2”), and phases (“Phase I/II”).

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

    sequential numbers and phases

    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
     ønsker
    -1.20
     verwacht
    -1.19
    bindet
    -1.16
     szükség
    -1.16
    でしたね
    -1.14
     to
    -1.11
    Ciri
    -1.10
     üç
    -1.09
     from
    -1.09
     that
    -1.09
    POSITIVE LOGITS
     is
    1.66
     natomiast
    1.19
     focuses
    1.09
     meanwhile
    1.09
     You
    1.09
     autre
    1.08
    は
    1.07
    主要是
    1.06
    ಢ
    1.03
    让你
    1.01
    Activations Density 0.031%

    No Known Activations