© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 67617
    Prev
    Next
    INDEX
    Explanations

    well known

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

    This neuron detects occurrences of boilerplate “well known in the art”–style phrases that appear in technical or patent-style writing.

    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
     właśnie
    -1.00
     algunas
    -0.97
     некоторых
    -0.95
     bazı
    -0.95
     algunos
    -0.93
     while
    -0.91
     some
    -0.91
     occasionally
    -0.91
     like
    -0.90
     only
    -0.89
    POSITIVE LOGITS
     already
    1.45
     вже
    1.41
    Already
    1.34
     již
    1.32
    already
    1.32
     уже
    1.32
     zaten
    1.30
     כבר
    1.30
    everybody
    1.29
     schon
    1.22
    Activations Density 0.081%

    No Known Activations