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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 5573
    Prev
    Next
    INDEX
    Explanations

    attends to the token "and" from tokens associated with specific competing pairs

    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
    0:0.10
    1:0.11
    2:0.08
    3:0.08
    4:0.10
    5:0.06
    6:0.17
    7:0.26
    Negative Logits
    DockStyle
    -0.34
    nino
    -0.32
    참고
    -0.29
    AndEndTag
    -0.29
     Elsie
    -0.28
    FOOTNOTES
    -0.28
    <tr>
    -0.27
    choso
    -0.27
     Viva
    -0.26
    chossen
    -0.26
    POSITIVE LOGITS
    Geplaatst
    0.40
    ."],
    0.39
     XCTest
    0.38
    .";
    
    0.38
    Datuak
    0.35
    )";
    
    0.35
    ).}
    0.34
     myſelf
    0.34
     متعلقه
    0.34
     referenties
    0.34
    Activations Density 0.018%

    No Known Activations