© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 130543
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly detects instances of the legal-boilerplate term “shall,” marking obligatory statements in formal texts.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    shall be or not

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    BSITE
    -2.17
    沨
    -2.06
     kreeg
    -1.96
    ⧠
    -1.96
     vulgaires
    -1.95
     mépris
    -1.91
     participado
    -1.90
    迥
    -1.88
     bahkan
    -1.86
     prepares
    -1.82
    POSITIVE LOGITS
    C
    3.05
    "
    2.72
    ).
    2.41
    {
    2.38
     be
    2.30
    ta
    2.19
    A
    2.17
    .”
    2.17
    \
    2.16
     to
    2.14
    Activations Density 0.004%

    No Known Activations