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 - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 1164
    Prev
    Next
    INDEX
    Explanations

    instances of "pg" followed by numeric or formatting symbols, indicating references to pages or terms in a scientific context

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

    references to experimental or study data

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

    PG and pgRNA

    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-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    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
    endregion
    -0.59
    HNO
    -0.56
     Wass
    -0.56
     ivelany
    -0.56
    WireFormatLite
    -0.56
     cdti
    -0.56
    textInput
    -0.56
     previsão
    -0.55
     ActionResult
    -0.54
    erski
    -0.54
    POSITIVE LOGITS
     PG
    0.91
    EndInit
    0.70
    NameInMap
    0.69
     pg
    0.66
    pgsql
    0.64
    PG
    0.63
     nahilalakip
    0.59
    GEBURTSDATUM
    0.58
     SLOT
    0.58
    mspace
    0.56
    Activations Density 0.038%

    No Known Activations