Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 1066
    Prev
    Next
    INDEX
    Explanations

    programming syntax elements and structure in code

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    syntax elements indicative of Java programming structure

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    code structure and syntax

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    fjspx
    -0.80
    writeFieldEnd
    -0.78
    SerializedSize
    -0.77
     للمعارف
    -0.74
    migrationBuilder
    -0.72
     الحره
    -0.72
     ويكيميديا
    -0.71
     ProtoMessage
    -0.70
     propOrder
    -0.66
    GOTREF
    -0.63
    POSITIVE LOGITS
    < 
    0.53
    >
    
    
    0.50
    ";
    
    0.48
     Aria
    0.47
    */
    
    
    0.47
    .";
    
    0.46
    );
    
    
    0.46
     grand
    0.46
     BaseModel
    0.45
    ();
    
    
    0.45
    Activations Density 0.019%

    No Known Activations