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. 2279
    Prev
    Next
    INDEX
    Explanations

    discussion of data analysis and statistical concepts

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

    Preceding tokens before technical terms

    np_acts-logits-general · gemini-2.0-flash

    creation and management

    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
     enfans
    -0.72
     varandra
    -0.63
     dieux
    -0.61
     mariée
    -0.61
     coû
    -0.60
     leçon
    -0.58
     épreuve
    -0.57
     définiti
    -0.57
     esegu
    -0.57
     marié
    -0.56
    POSITIVE LOGITS
     creation
    1.16
     management
    1.03
     acquisition
    1.03
     removal
    1.00
     formation
    0.99
     development
    0.95
     selection
    0.94
     preparation
    0.92
     analysis
    0.91
     evaluation
    0.89
    Activations Density 1.603%

    No Known Activations