© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 72241
    Prev
    Next
    INDEX
    Explanations

    when X or Y

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

    The neuron selectively activates on words that appear as part of document titles or section headings (especially the first few content words of a headline).

    oai_token-act-pair · o4-miniTriggered 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
    Not in Any Lists

    No Comments

    Negative Logits
     horrend
    -1.30
     insanely
    -1.27
     craz
    -1.23
    8
    -1.23
     horrid
    -1.23
     horrifying
    -1.22
     revamped
    -1.21
     compatri
    -1.20
     hugely
    -1.18
     laço
    -1.14
    POSITIVE LOGITS
     also
    1.48
     gick
    1.37
     just
    1.32
     both
    1.30
     an
    1.29
     what
    1.28
     as
    1.27
     fanns
    1.27
    )।
    1.25
     Singapur
    1.25
    Activations Density 0.139%

    No Known Activations