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

    Neuronpedia

    APICircuit 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. 0-GEMMASCOPE-ATT-16K
    6. 72
    Prev
    Next
    INDEX
    Explanations

    attends to various tokens marked as "a" from paired tokens that denote the title or name of something, which suggests a focus on identifying or introducing specific items or concepts

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k
    Configuration
    google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.0.attn.hook_z
    Hook Layer
    0
    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
    0:0.03
    1:0.03
    2:0.02
    3:0.02
    4:0.22
    5:0.59
    6:0.02
    7:0.02
    Negative Logits
     the
    -0.96
     an
    -0.69
     a
    -0.68
     one
    -0.63
     A
    -0.59
     
    -0.53
     of
    -0.52
     N
    -0.51
     S
    -0.51
    </b>
    -0.50
    POSITIVE LOGITS
     itſelf
    1.07
     Theſe
    0.96
     uſed
    0.95
     myſelf
    0.94
     themſelves
    0.94
     auffi
    0.94
     ſche
    0.93
     Monfieur
    0.91
     leſs
    0.91
     raiſ
    0.91
    Activations Density 6.091%

    No Known Activations