© 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. 33119
    Prev
    Next
    INDEX
    Explanations

    legal terms and concepts

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

    The neuron fires on statutory or regulatory language—specifically on modal/legal‐obligation terms (e.g. “shall,” “license,” “search,” “seize,” “sale”) in legal statutes.

    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
     only
    -1.37
     many
    -1.30
     fós
    -1.20
     münchen
    -1.17
     colega
    -1.16
     aspects
    -1.16
     numerous
    -1.15
     चीज़ों
    -1.15
     dolay
    -1.14
     néhány
    -1.14
    POSITIVE LOGITS
    alians
    1.32
     السياس
    1.31
     الحكومة
    1.28
    seekers
    1.20
    []){
    1.20
     blackish
    1.17
    ിൽ
    1.16
    athered
    1.15
    righ
    1.14
     hugely
    1.14
    Activations Density 0.006%

    No Known Activations