Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 5232
    Prev
    Next
    INDEX
    Explanations

    occurrences of the term "Mar" followed by relevant numerical and descriptive modifiers

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    starting with "Mar"

    np_acts-logits-general · gemini-2.0-flash

    Mar names and places

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    kegaard
    -0.50
    WindowConstants
    -0.49
     wire
    -0.47
     FJ
    -0.46
    RRect
    -0.45
     Ach
    -0.45
    invokeLater
    -0.45
    Ұ
    -0.45
    🥞
    -0.45
    сылкі
    -0.45
    POSITIVE LOGITS
     ExecuteAsync
    0.74
    migrationBuilder
    0.73
     Madness
    0.69
     AspNetCore
    0.68
     Marr
    0.68
    
    0.67
     Marl
    0.66
     MAR
    0.63
     insuffisamment
    0.62
    TintMode
    0.62
    Activations Density 0.140%

    No Known Activations