© 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. 22-GEMMASCOPE-RES-131K
    7. 127980
    Prev
    Next
    INDEX
    Explanations

    The neuron is picking out “code‐style” identifier tokens—things like version numbers, module or domain names, and other tokens containing digits or uppercase letters.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    software development and files

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    衹
    -1.07
     Gastgeber
    -1.06
     betrek
    -1.02
     deoarece
    -0.99
     acestui
    -0.98
    澆
    -0.96
     împreună
    -0.96
     píše
    -0.94
    うまい
    -0.94
    也會
    -0.94
    POSITIVE LOGITS
    alties
    1.23
     espectáculos
    1.20
    Elő
    1.10
    繭
    1.09
    २
    1.08
     FileWriter
    1.04
     higiene
    1.04
     pages
    0.98
    drawSprites
    0.96
     icmp
    0.96
    Activations Density 0.002%

    No Known Activations