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

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 5203
    Prev
    Next
    INDEX
    Explanations

    specific quantities or occurrences of items and their relationships in contexts such as legal rights, measurements, or culinary instructions

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

    Preceding certain words (of, BL, obese, cupcake)

    np_acts-logits-general · gemini-2.0-flash

    each or every

    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
     manifolds
    -0.57
     serials
    -0.56
     wickets
    -0.56
     basses
    -0.55
     pitchers
    -0.55
     assholes
    -0.54
     pianos
    -0.54
     embodiments
    -0.54
    providedIn
    -0.54
     aprons
    -0.53
    POSITIVE LOGITS
     bienfaits
    0.95
     âmes
    0.95
     pouvoirs
    0.94
     vœux
    0.94
     négociations
    0.93
     dieux
    0.90
     bénéfices
    0.90
     prochaines
    0.87
     vérit
    0.87
     Grecs
    0.86
    Activations Density 2.905%

    No Known Activations