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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Attention Out - 16k
    5. 10-GEMMASCOPE-ATT-16K
    6. 6409
    Prev
    Next
    INDEX
    Explanations

    attends to specific isolated or quantified tokens from a range of more general or category tokens

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k
    Configuration
    google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.10.attn.hook_z
    Hook Layer
    10
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.23
    1:0.24
    2:0.10
    3:0.07
    4:0.06
    5:0.03
    6:0.09
    7:0.15
    Negative Logits
     voeten
    -0.42
    <tr>
    -0.40
    flikt
    -0.40
     étrangère
    -0.39
     réelle
    -0.39
    Ruff
    -0.38
     Cushman
    -0.38
     avancée
    -0.38
    opensource
    -0.38
    CodeDom
    -0.38
    POSITIVE LOGITS
    )";
    
    0.77
     ModelExpression
    0.74
    )");
    
    0.73
    ();)
    0.69
    `;
    
    0.65
    "]);
    
    0.65
    .";
    
    0.63
    ']);
    
    0.63
    "])
    
    0.63
    %)$
    0.62
    Activations Density 30.139%

    No Known Activations