© 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. 10-GEMMASCOPE-RES-131K
    4. 120079
    Prev
    Next
    INDEX
    Explanations

    reduce or remove with cut

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

    The neuron detects occurrences of “cut” (and its variants like “cutting,” “cut down,” “cut out,” “cut back,” etc.) indicating some form of trimming or reduction.

    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
     Allowing
    -1.74
    n
    -1.73
     a
    -1.70
    
    -1.62
    (
    -1.60
    你会
    -1.59
     two
    -1.59
    Источники
    -1.57
     unwise
    -1.55
     Acknowled
    -1.50
    POSITIVE LOGITS
    蚴
    1.96
    </b>
    1.95
     églises
    1.77
    ),
    1.76
    茕
    1.70
     []
    
    1.69
     ""))
    1.69
    訐
    1.68
    籴
    1.67
     '/';
    1.66
    Activations Density 0.010%

    No Known Activations