© 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. 64566
    Prev
    Next
    INDEX
    Explanations

    programming literals and types

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

    The neuron strongly activates on programming-language type tokens (e.g. “string”, “REAL”, “xsd”) when they appear in code or schema definitions.

    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
     Since
    -1.34
     nine
    -1.29
     only
    -1.28
     On
    -1.27
     D
    -1.27
     In
    -1.26
     on
    -1.25
     F
    -1.24
     creates
    -1.23
     by
    -1.23
    POSITIVE LOGITS
    "+
    1.41
    ,
    1.39
     \
    1.32
     réinitial
    1.27
     "
    1.26
    quelles
    1.25
    ,\\
    1.25
     
    1.25
     dieß
    1.24
     longterm
    1.23
    Activations Density 0.145%

    No Known Activations