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. 10-GEMMASCOPE-ATT-16K
    6. 5928
    Prev
    Next
    INDEX
    Explanations

    attends to the token "Errorf" from tokens containing punctuation or structural markers

    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
    0:0.09
    1:0.11
    2:0.10
    3:0.08
    4:0.07
    5:0.06
    6:0.17
    7:0.30
    Negative Logits
    šanās
    -0.23
     Clearly
    -0.22
    Clearly
    -0.22
    šanu
    -0.22
     sink
    -0.21
    nitte
    -0.21
    预
    -0.20
     DPRD
    -0.20
     sunk
    -0.20
    нятно
    -0.20
    POSITIVE LOGITS
     Normdatei
    0.52
     препратки
    0.49
     ExecuteAsync
    0.44
     censiti
    0.43
    hyrchwyd
    0.42
    findpost
    0.40
    MethodManager
    0.40
    enderror
    0.40
    ]="
    0.39
    asteroide
    0.38
    Activations Density 0.020%

    No Known Activations