© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 14484
    Prev
    Next
    INDEX
    Explanations

    code execution errors

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

    The neuron fires on tokens that occur in programming error messages—especially those about calling or accessing invalid members, functions, methods, or properties.

    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
    sqcup
    -1.02
     with
    -1.02
     arba
    -1.02
     during
    -0.96
     through
    -0.93
     if
    -0.93
     متحده
    -0.92
    
    -0.91
     If
    -0.90
    Suara
    -0.90
    POSITIVE LOGITS
     romero
    0.98
    quickly
    0.97
     accéder
    0.95
     lác
    0.94
     raso
    0.93
     recuperação
    0.93
     passos
    0.92
     chè
    0.91
     inond
    0.91
     ginge
    0.91
    Activations Density 0.030%

    No Known Activations