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. 6394
    Prev
    Next
    INDEX
    Explanations

    attends to the "%" token from URL-related 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
    0:0.07
    1:0.18
    2:0.08
    3:0.07
    4:0.13
    5:0.15
    6:0.12
    7:0.14
    Negative Logits
     whoſe
    -0.42
     Monfieur
    -0.38
     ſte
    -0.36
     raiſ
    -0.36
     itſelf
    -0.36
     linkovi
    -0.36
     Jefus
    -0.35
    APOLIS
    -0.35
     myſelf
    -0.35
    ientras
    -0.35
    POSITIVE LOGITS
    </em>
    0.28
     birth
    0.26
    brat
    0.25
     Мат
    0.24
     letz
    0.23
     left
    0.23
    </h2>
    0.23
     *
    0.23
    InvalidProtocol
    0.23
    most
    0.23
    Activations Density 0.014%

    No Known Activations