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. 6529
    Prev
    Next
    INDEX
    Explanations

    attends to specific programming keywords or symbols from generic 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
    0:0.06
    1:0.07
    2:0.07
    3:0.08
    4:0.47
    5:0.07
    6:0.06
    7:0.07
    Negative Logits
     Gell
    -0.38
    RenderAtEndOf
    -0.36
    TagMode
    -0.35
    ьаж
    -0.35
    ParallelGroup
    -0.34
     <=",
    -0.34
    -------------</
    -0.32
    
    -0.31
     שוליים
    -0.31
    ">//
    -0.30
    POSITIVE LOGITS
     TAMBIÉN
    0.37
     firebaseConfig
    0.32
    brigens
    0.31
     NDEBUG
    0.31
    ApplicationTests
    0.31
     useDispatch
    0.31
     requestId
    0.31
    bish
    0.30
     دیکھیے
    0.30
     Também
    0.30
    Activations Density 0.379%

    No Known Activations