© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 127955
    Prev
    Next
    INDEX
    Explanations

    This neuron activates on nested or self-referential structures—e.g. repeated or chained occurrences of the same token (like “near‐near‐infra‐red”), nested function calls (parent().parent), or nested arrays.

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

    data structures and references

    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
     fewer
    -0.97
     while
    -0.91
     ISSUED
    -0.89
    んでしょう
    -0.88
    mergeFrom
    -0.87
     waarom
    -0.83
     either
    -0.83
     telev
    -0.83
    bá
    -0.83
    latimes
    -0.82
    POSITIVE LOGITS
     of
    1.24
     original
    0.99
    ons
    0.97
    original
    0.95
    w
    0.92
     originaux
    0.92
    f
    0.91
    !
    0.91
    待つ
    0.90
    artment
    0.89
    Activations Density 0.044%

    No Known Activations