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. 6374
    Prev
    Next
    INDEX
    Explanations

    attends to specific numerical values followed by characters or symbols from references or additional notations

    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
    0:0.12
    1:0.13
    2:0.14
    3:0.10
    4:0.13
    5:0.10
    6:0.10
    7:0.16
    Negative Logits
    ISupport
    -0.34
    DoubleQuotes
    -0.29
    nestjs
    -0.26
    hyrchwyd
    -0.25
    "
    -0.23
    devServer
    -0.23
     drugiej
    -0.23
    ガル
    -0.22
     useStyles
    -0.22
    U
    -0.22
    POSITIVE LOGITS
     Мексичка
    0.42
     actionMode
    0.37
    ();)
    0.34
    '],$
    0.32
     MenuView
    0.31
    )";
    
    0.31
    martre
    0.30
    lück
    0.30
    ,:),
    0.29
     Réponses
    0.29
    Activations Density 0.810%

    No Known Activations