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. 1209
    Prev
    Next
    INDEX
    Explanations

    attends to blank or empty tokens from various preceding tokens in a sequence

    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
    0:0.04
    1:0.44
    2:0.02
    3:0.01
    4:0.01
    5:0.01
    6:0.09
    7:0.03
    8:0.03
    9:0.04
    10:0.06
    11:0.01
    12:0.02
    13:0.04
    14:0.05
    15:0.05
    Negative Logits
    <pad>
    -0.98
    <unused3>
    -0.97
    <unused17>
    -0.97
    <unused52>
    -0.97
    <unused68>
    -0.97
    [@BOS@]
    -0.97
    <unused8>
    -0.97
    <unused51>
    -0.97
    <unused79>
    -0.97
    <unused43>
    -0.97
    POSITIVE LOGITS
     nocturno
    0.43
     prohibido
    0.39
     húmedo
    0.39
     esclavos
    0.38
     fotográfico
    0.38
     orgánico
    0.36
     extranjera
    0.36
     Económica
    0.36
     muerta
    0.35
     rodillas
    0.35
    Activations Density 0.069%

    No Known Activations