© 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. 129336
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on code‐specific tokens—especially punctuation (colons, parentheses, ampersands) and programming identifiers (type names, numeric literals)—marking snippets of source code.

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

    This neuron strongly activates on code identifiers—especially type names, constructors, macros, and other programming-style tokens.

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

    code and configuration settings

    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
    ؤول
    -0.99
    tellungs
    -0.88
     fibras
    -0.86
    てて
    -0.83
     Suecia
    -0.83
     meistens
    -0.82
    İyi
    -0.82
    Jährige
    -0.81
     kennel
    -0.81
    λους
    -0.80
    POSITIVE LOGITS
        
    1.15
         
    1.11
       
    1.05
             
    1.04
          
    0.96
           
    0.94
              
    0.93
                    
    0.90
                      
    0.90
                              
    0.89
    Activations Density 0.211%

    No Known Activations