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. 1197
    Prev
    Next
    INDEX
    Explanations

    attends to the token "0" from various numerical tokens

    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
    0:0.13
    1:0.06
    2:0.02
    3:0.07
    4:0.09
    5:0.05
    6:0.10
    7:0.06
    8:0.02
    9:0.03
    10:0.03
    11:0.07
    12:0.09
    13:0.05
    14:0.02
    15:0.05
    Negative Logits
     queſta
    -2.56
    <unused68>
    -2.55
    <unused52>
    -2.55
    [@BOS@]
    -2.53
    <unused79>
    -2.53
    <unused14>
    -2.53
    <unused28>
    -2.53
    <unused16>
    -2.53
    <unused3>
    -2.53
    <unused8>
    -2.53
    POSITIVE LOGITS
    3
    1.09
    2
    1.09
    0
    1.05
    4
    0.99
    1
    0.98
    6
    0.98
    5
    0.96
    7
    0.91
    8
    0.91
    9
    0.90
    Activations Density 0.049%

    No Known Activations