© 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. 25710
    Prev
    Next
    INDEX
    Explanations

    sequences of words

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

    The neuron fires on the token “comment” (as in comment IDs or links), i.e. it’s detecting comment‐related metadata.

    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
     whimsical
    -1.21
    ্যান
    -1.09
     wondrous
    -1.07
     Además
    -1.04
    歴代
    -1.02
     adjunto
    -1.02
    さみ
    -0.98
     LITER
    -0.97
     farmacé
    -0.95
     vários
    -0.95
    POSITIVE LOGITS
     new
    1.38
     on
    1.13
     each
    1.11
     five
    1.09
    dillera
    1.05
    ！」
    1.05
     one
    1.02
     after
    1.02
     eight
    1.02
     that
    1.02
    Activations Density 0.000%

    No Known Activations