© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 129472
    Prev
    Next
    INDEX
    Explanations

    This neuron detects references to source‐code line numbers in error or warning messages (e.g. the “on line N” pattern).

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

    code, error, or comment markers

    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
     if
    -1.88
     one
    -1.68
     because
    -1.67
     what
    -1.61
     on
    -1.59
     Creating
    -1.44
     meticulously
    -1.42
     should
    -1.39
     in
    -1.38
     for
    -1.38
    POSITIVE LOGITS
     Dinas
    1.58
     AMAZING
    1.43
     horrid
    1.39
     lâng
    1.38
    
    1.36
     boister
    1.36
     wondrous
    1.35
    
    1.34
     specif
    1.34
     écouter
    1.32
    Activations Density 0.000%

    No Known Activations