© 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. 73737
    Prev
    Next
    INDEX
    Explanations

    PRECEDENTIAL, NO

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

    The neuron strongly activates on fully capitalized tokens and punctuation used in document headings (e.g. section titles, quotes, and other all-caps header text).

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408

    whitespace and formatting characters in legal or formal documents, particularly tabs and spaces used for indentation and alignment.

    oai_token-act-pair · claude-4-5-sonnet
    Triggered by @jyhe0408

    formal document headers and typographical markers, especially all-caps legal headings, docket numbers, and opening quotation punctuation.

    oai_token-act-pair · gpt-5
    Triggered 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
    Negative Logits
     although
    -0.90
     there
    -0.81
    ousness
    -0.81
     Since
    -0.79
    قى
    -0.79
     Although
    -0.79
     somethin
    -0.79
    اعي
    -0.79
     makes
    -0.78
     <<<
    -0.78
    POSITIVE LOGITS
    sions
    0.93
     très
    0.89
    pent
    0.88
    gage
    0.82
    pada
    0.81
    clean
    0.80
     product
    0.79
    sn
    0.79
    titutions
    0.78
    show
    0.78
    Activations Density 0.036%

    No Known Activations