© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 130546
    Prev
    Next
    INDEX
    Explanations

    source followed by types

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

    The neuron primarily activates on occurrences of the word “source” (in any casing) in code—e.g., in macros like `#define SOURCE_DIR` and in comment headers.

    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
    I
    -3.55
    -
    -3.08
    [
    -2.94
    if
    -2.89
    _
    -2.84
    and
    -2.78
    ти
    -2.61
    of
    -2.50
    ',
    -2.45
    n
    -2.38
    POSITIVE LOGITS
    
    2.95
    뻣
    2.84
    趼
    2.81
    ￮
    2.81
     \
    2.77
    咵
    2.75
    ⚌
    2.69
    
    2.69
     hembra
    2.66
    
    2.66
    Activations Density 0.048%

    No Known Activations