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. Residual Stream - 65k
    5. 17-GEMMASCOPE-RES-65K
    6. 28481
    Prev
    Next
    INDEX
    Explanations

    the presence of specific tokens used for indicating the start of a new section or text block

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.17.hook_resid_post
    Hook Layer
    17
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     wszystkich
    -0.75
     kaikki
    -0.73
    MigrationBuilder
    -0.73
     všech
    -0.72
     everytime
    -0.70
    everything
    -0.69
     každé
    -0.68
    所有人都
    -0.68
    everyone
    -0.68
     všetkých
    -0.67
    POSITIVE LOGITS
    ই
    0.68
    ństw
    0.59
    enumii
    0.59
    XmlAccessType
    0.58
     now
    0.58
    άλ
    0.58
     either
    0.57
    Either
    0.57
     Either
    0.56
    甚至是
    0.56
    Activations Density 0.115%

    No Known Activations