© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 113888
    Prev
    Next
    INDEX
    Explanations

    Enron emails and technical terms

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

    technical or programming-related terms, particularly in code, technical documentation, or formal syntax.

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

    The neuron responds to uncommon, specialized tokens—often proper names or domain‐specific terms—regardless of context.

    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
     própri
    -1.55
     robuste
    -1.50
     menjelaskan
    -1.39
     rése
    -1.39
     regering
    -1.37
     parfü
    -1.37
    uldades
    -1.34
     mengatakan
    -1.34
    “[
    -1.33
     rengi
    -1.32
    POSITIVE LOGITS
     at
    1.76
     by
    1.75
     like
    1.64
     one
    1.62
     only
    1.47
     '
    1.28
     about
    1.27
     either
    1.27
     just
    1.27
     actually
    1.26
    Activations Density 0.314%

    No Known Activations