© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 43852
    Prev
    Next
    INDEX
    Explanations

    acylindrical, robocopy command, work of

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

    The neuron strongly activates on compound words made by joining two smaller words (e.g. criss + cross, robo + copy, some + thing).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     where
    -1.77
     need
    -1.75
     own
    -1.65
    EN
    -1.52
     odpowied
    -1.48
    ordine
    -1.46
     in
    -1.45
     another
    -1.44
     tell
    -1.43
     as
    -1.42
    POSITIVE LOGITS
     judicia
    1.58
     fraî
    1.57
     plastiques
    1.54
     idyl
    1.48
     conster
    1.48
     både
    1.47
     marte
    1.47
     fuis
    1.46
     raiſ
    1.45
     boute
    1.45
    Activations Density 0.046%

    No Known Activations