© 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. 105568
    Prev
    Next
    INDEX
    Explanations

    metadata labels and values

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

    structured metadata fields and formatting elements like colons, newlines, and field labels in formal documents.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron is specialized to spot structured metadata labels and their separators (e.g. “Genre:”, “Location:”, “Author:”, etc.) in document headers.

    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
     &
    -1.70
    the
    -1.67
    来到了
    -1.44
    The
    -1.43
    1
    -1.41
    -1.38
    &
    -1.38
    {
    -1.35
    $
    -1.35
     zostać
    -1.28
    POSITIVE LOGITS
    Industrie
    1.54
     bija
    1.48
    裏面
    1.46
    larla
    1.45
     kanssa
    1.40
     ktoré
    1.40
    concep
    1.39
     ABBRE
    1.34
     které
    1.32
    ᐟ
    1.31
    Activations Density 0.035%

    No Known Activations