© 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. 10-GEMMASCOPE-RES-131K
    7. 99192
    Prev
    Next
    INDEX
    Explanations

    legal and contractual language

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

    The neuron fires on tokens that mark structural or metadata elements in legal or technical texts—things like section or exhibit numbers, citations (e.g. “U.S. Pat. No.”), headings, and possessive endings.

    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
     الجلد
    -1.61
     botanique
    -1.59
     papillon
    -1.56
     comte
    -1.56
     poulet
    -1.56
     africain
    -1.54
     pernic
    -1.48
     frappé
    -1.47
     herbes
    -1.45
     eccl
    -1.45
    POSITIVE LOGITS
     your
    2.64
     if
    2.13
     or
    2.03
     either
    2.00
     by
    1.84
     them
    1.68
     necessary
    1.65
     required
    1.57
     unless
    1.56
     described
    1.55
    Activations Density 0.507%

    No Known Activations