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 Post
    5. 12-GEMMASCOPE-RES-1M
    6. 502
    Prev
    Next
    INDEX
    Explanations

    concepts related to clever solutions and their effectiveness in various systems or architectures

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

    clever

    np_max-act · gemini-2.0-flashTriggered by @johnny
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-res/layer_12/width_1m/average_l0_107
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    1,048,576
    Data Type
    float32
    Hook Name
    blocks.12.hook_resid_post
    Hook Layer
    12
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    sizeCache
    -0.81
     occasionally
    -0.60
    WireFormatLite
    -0.60
    RefNanny
    -0.59
     Occasionally
    -0.56
    Occasionally
    -0.53
    Билгалдахарш
    -0.53
     sometimes
    -0.52
     recently
    -0.50
     frequently
    -0.49
    POSITIVE LOGITS
     بيها
    0.62
     sicherzustellen
    0.57
     garanzia
    0.57
     Público
    0.56
    IBU
    0.55
    roek
    0.54
    RoutingModule
    0.54
     Aurang
    0.53
    ügel
    0.52
     bruto
    0.52
    Activations Density 0.017%

    No Known Activations