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 - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 2416
    Prev
    Next
    INDEX
    Explanations

    specific elements and structures commonly found in programming languages, particularly relating to variable assignments and data types

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

    Digits followed by other characters

    np_acts-logits-general · gemini-2.0-flash

    binary and hexademical digits

    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
    ftagPool
    -0.55
    RTHOOK
    -0.54
    Inches
    -0.52
    Vinyl
    -0.49
     formulation
    -0.49
    ")){
    
    -0.48
    ([]);
    -0.47
    chas
    -0.47
     six
    -0.47
    addContainerGap
    -0.47
    POSITIVE LOGITS
     Roskov
    0.71
    ItemBackground
    0.61
     Meksiku
    0.60
     Мексичка
    0.59
     Савезне
    0.58
    LookAnd
    0.58
     doubl
    0.57
     vectorstock
    0.56
     벡터
    0.56
     متحده
    0.56
    Activations Density 0.382%

    No Known Activations