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. 6447
    Prev
    Next
    INDEX
    Explanations

    attends to instances of the token "break" from instances of "break" that are further in the sequence

    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
    1:0.13
    2:0.14
    3:0.04
    4:0.15
    5:0.26
    6:0.07
    7:0.07
    Negative Logits
     potere
    -0.27
     щодо
    -0.26
    ič
    -0.25
    rinth
    -0.24
     alcun
    -0.24
    まで
    -0.23
     بأن
    -0.23
     cargas
    -0.23
     un
    -0.23
    شتی
    -0.23
    POSITIVE LOGITS
     виправивши
    0.44
     getItemId
    0.40
    "])
    
    0.40
    setVerticalGroup
    0.39
    mapStateToProps
    0.39
     onCancelled
    0.39
    "],
    
    0.38
     sẻ
    0.38
    __':
    
    0.37
    AnchorStyles
    0.37
    Activations Density 0.003%

    No Known Activations