© 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. 10-GEMMASCOPE-RES-131K
    7. 68389
    Prev
    Next
    INDEX
    Explanations

    NLS Respondent-Appellee

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

    The neuron fires on blocks of ASCII-art–style punctuation (long runs of underscores, slashes, backslashes, pipes, etc.) used as decorative separators or illustrations.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     like
    -1.74
     it
    -1.68
     both
    -1.65
     itself
    -1.58
     using
    -1.53
     you
    -1.52
     these
    -1.52
     tells
    -1.48
     how
    -1.48
     contain
    -1.45
    POSITIVE LOGITS
    takings
    1.52
    ceased
    1.52
     herbes
    1.49
     probablement
    1.44
     trône
    1.43
     applau
    1.42
    Evet
    1.41
     superbe
    1.40
    Dieu
    1.40
     sembl
    1.39
    Activations Density 0.009%

    No Known Activations