© 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. 72815
    Prev
    Next
    INDEX
    Explanations

    page numbers or section markers

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

    This neuron picks out boilerplate/structural tokens in web or formatted text—things like isolated numbers or section markers, HTML tags or prompts (e.g. “p” tags, “Ask HN:”), URLs (“http”), and navigation cues (“Read More →”).

    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
     can
    -1.05
     not
    -1.04
    Koordinaten
    -1.02
    after
    -1.02
     Get
    -1.00
    get
    -0.99
     get
    -0.96
     confirmar
    -0.96
     как
    -0.96
     How
    -0.95
    POSITIVE LOGITS
     feltro
    1.18
     burbujas
    1.18
    iog
    1.15
     riego
    1.07
    inspiring
    1.05
     cór
    1.03
     šaty
    1.03
    poop
    1.02
     Outro
    1.01
    WASH
    0.99
    Activations Density 0.009%

    No Known Activations