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-2-9B
    4. Attention Out - 16k
    5. 0-GEMMASCOPE-ATT-16K
    6. 1269
    Prev
    Next
    INDEX
    Explanations

    attends to the token "living" from various locations or conditions marked by related prepositions

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-9b-pt-att/layer_0/width_16k/average_l0_61
    Prompts (Dashboard)
    16,384 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
    0:0.01
    1:0.02
    2:0.51
    3:0.02
    4:0.00
    5:0.00
    6:0.07
    7:0.04
    8:0.00
    9:0.00
    10:0.00
    11:0.01
    12:0.00
    13:0.14
    14:0.08
    15:0.00
    Negative Logits
    ’
    -0.29
    '
    -0.25
    c
    -0.22
    ;
    -0.22
     künftig
    -0.20
    .
    -0.20
    :
    -0.20
    inf
    -0.20
    mathrm
    -0.20
    left
    -0.19
    POSITIVE LOGITS
    OGND
    0.57
     Roskov
    0.57
     betweenstory
    0.57
     kasarigan
    0.56
    IndentedString
    0.55
    <unused52>
    0.54
    <unused8>
    0.54
    <unused47>
    0.54
    <unused3>
    0.53
    <unused14>
    0.53
    Activations Density 0.035%

    No Known Activations