© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 71984
    Prev
    Next
    INDEX
    Explanations

    profession

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

    The neuron is picking up on the little “figure‐space”/formatting characters that mark metadata or layout (the special whitespace tokens around dates, bylines, section headers, etc.). In other words, it fires on those extra spacing characters used to format things like datelines, headings, and other structural bits of the 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_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
     versuchen
    -0.94
     سبورت
    -0.91
    că
    -0.85
    katu
    -0.83
     мужские
    -0.83
     everything
    -0.82
     amanecer
    -0.82
     creative
    -0.80
    VED
    -0.79
     Financial
    -0.78
    POSITIVE LOGITS
    懂得
    0.91
     fundamentos
    0.87
    ꜜ
    0.82
     orale
    0.81
    Mains
    0.81
    裄
    0.80
    égal
    0.80
    snowy
    0.78
     professional
    0.77
    ᴱ
    0.77
    Activations Density 0.142%

    No Known Activations