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. 6402
    Prev
    Next
    INDEX
    Explanations

    attends to array-related tokens from list-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
    1:0.09
    2:0.08
    3:0.46
    4:0.06
    5:0.03
    6:0.08
    7:0.08
    Negative Logits
    lação
    -0.34
    modules
    -0.32
     entire
    -0.31
    sitis
    -0.31
    lob
    -0.30
    IInterface
    -0.30
    Darth
    -0.30
    toff
    -0.30
    rino
    -0.30
     whole
    -0.29
    POSITIVE LOGITS
     EconPapers
    0.52
     typelib
    0.45
    )";
    
    0.41
    ConstraintMaker
    0.39
     disponibilités
    0.39
    IntoConstraints
    0.38
    IsMutable
    0.37
     varandra
    0.37
    Geplaatst
    0.36
     validamos
    0.36
    Activations Density 0.663%

    No Known Activations