Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 5522
    Prev
    Next
    INDEX
    Explanations

    attends to the tokens "either" and "or" from "both" and "and" 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
    0:0.11
    1:0.11
    2:0.54
    3:0.04
    4:0.03
    5:0.02
    6:0.02
    7:0.08
    Negative Logits
    وصلات
    -0.40
    ,:);
    -0.38
    ampton
    -0.38
     <>",
    -0.36
    HasAnnotation
    -0.36
     estekak
    -0.35
    dotenv
    -0.35
    achella
    -0.34
    umsy
    -0.34
     Palmas
    -0.34
    POSITIVE LOGITS
     pinulongan
    0.38
     internazionali
    0.33
    AnchorTagHelper
    0.31
     lourdes
    0.31
     mengajar
    0.30
     asiatique
    0.30
     CURIAM
    0.30
     internationaux
    0.30
     eneste
    0.30
     Вес
    0.30
    Activations Density 0.627%

    No Known Activations