© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 100756
    Prev
    Next
    INDEX
    Explanations

    code and programming constructs

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

    underscores or commas used as placeholder variables in iteration loops (for/foreach statements) across different programming languages.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron fires on iteration constructs in code—tokens like “for,” “BOOST_FOREACH,” “ng-repeat,” placeholders in loops (“_,”) and related separators that signal a loop.

    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
    穑
    -2.05
    砗
    -1.80
     memutus
    -1.77
     если
    -1.73
     jis
    -1.72
     abbastanza
    -1.70
    魴
    -1.69
    
    -1.69
     ')
    -1.67
     soms
    -1.67
    POSITIVE LOGITS
    寕
    1.84
     who
    1.83
     become
    1.70
     setelah
    1.69
     such
    1.66
    تیجه
    1.64
    大半
    1.63
    颪
    1.55
    pecabe
    1.54
    鑷
    1.53
    Activations Density 0.001%

    No Known Activations