© 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
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 62642
    Prev
    Next
    INDEX
    Explanations

    YouTube series, weekly ad, dance cover

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

    The neuron fires on structured “metadata” tokens in informational summaries—things like birth dates (“born”), nationality (“American”), attribution phrases (“best known for,” “his YouTube”), numeric identifiers and counts (years, view-counts, DOI numbers, percentages), and section-style labels (“Weekly,” “Tour,” “Dance Cover”).

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Negative Logits
     our
    -0.97
     five
    -0.94
     at
    -0.92
     without
    -0.91
     both
    -0.91
     some
    -0.90
     their
    -0.89
     actions
    -0.89
     any
    -0.87
     during
    -0.86
    POSITIVE LOGITS
    ཥ
    1.02
     sier
    0.99
    legt
    0.98
     Danh
    0.98
     alho
    0.97
    ⁸
    0.96
     maquilla
    0.94
    isiin
    0.94
    ztes
    0.94
     kanker
    0.93
    Activations Density 0.009%

    No Known Activations