© 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. 34-GEMMASCOPE-RES-131K
    7. 23561
    Prev
    Next
    INDEX
    Explanations

    group address

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

    The neuron is sensitive to tokens that serve as structural or metadata markers—like section headings, titles, episode/segment numbers, or other line‐initial labels—rather than regular running text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    бенок
    -1.24
    の方に
    -0.75
     plát
    -0.74
    varing
    -0.73
    prech
    -0.72
     منذ
    -0.72
    SIGINT
    -0.71
     escolas
    -0.71
    mutter
    -0.71
     тебя
    -0.70
    POSITIVE LOGITS
     guys
    5.66
     folks
    4.13
    Guys
    3.81
     ladies
    3.80
     gentlemen
    3.69
     Guys
    3.67
    guys
    3.63
     lads
    3.53
     fellas
    3.47
     boys
    3.23
    Activations Density 0.197%

    No Known Activations