© 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. 69120
    Prev
    Next
    INDEX
    Explanations

    confirming details before action

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

    The neuron fires on directive or checklist language in instructional or warning texts—terms like “review,” “verify,” “ensure,” “follow,” etc., that tell the reader to check or confirm something.

    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
    Karakter
    -0.94
    iveau
    -0.91
    Klas
    -0.89
     benzin
    -0.87
    ebenarnya
    -0.85
     attempts
    -0.85
    ჩ
    -0.85
     nabo
    -0.81
    ulitan
    -0.80
     Argentino
    -0.79
    POSITIVE LOGITS
     before
    1.20
    确认
    1.03
     sebelum
    0.99
    决定
    0.90
    IZONA
    0.84
     soprattutto
    0.84
     notamment
    0.84
    final
    0.82
     verkligen
    0.82
     confirm
    0.82
    Activations Density 0.025%

    No Known Activations