© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 28356
    Prev
    Next
    INDEX
    Explanations

    predicting word endings

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

    The neuron is highly activated by imperative or call-to-action verbs (like “write” or “contribute”) that urge the reader to perform an action.

    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
     all
    -1.21
     now
    -1.10
     perí
    -1.09
     always
    -1.09
     several
    -1.09
     blatantly
    -1.07
     begin
    -1.07
     purposefully
    -1.05
     purposely
    -1.05
     virtually
    -1.05
    POSITIVE LOGITS
    wherein
    1.16
     hermosas
    1.09
     ___________
    1.08
    ith
    1.05
    often
    1.04
     dificultades
    1.04
     ____________
    1.04
     _________
    1.03
     EIGHT
    1.03
     HAVING
    1.02
    Activations Density 0.000%

    No Known Activations