© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 16k
    9. 20-GEMMASCOPE-RES-16K
    11. 9289
    Prev
    Next
    INDEX
    Explanations

    references to data structures or coding concepts

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

    technical terms, identifiers, and structured data elements (such as code components, variable names, database fields, key-value pairs, chemical compounds, and format-specific elements) in technical or data-driven contexts.

    oai_token-act-pair · gpt-4.1-2025-04-14Triggered by @ywhp2c8tjm

    Followed by a description

    np_acts-logits-general · gemini-2.0-flash

    proper noun or technical terms

    np_acts-logits-general · gemini-2.5-flash-lite

    domain-specific, content‑rich nouns and technical or formal terminology (including proper names and jargon).

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    Prompts (Dashboard)
    24,576 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
     ویکی‌پدی
    -1.16
    EndGlobalSection
    -1.00
    featureID
    -0.99
     Administrativna
    -0.96
    ьаж
    -0.94
     noDo
    -0.89
    msgTypes
    -0.88
    IndentedString
    -0.83
    ंदीखरीदारी
    -0.79
    contentLoaded
    -0.77
    POSITIVE LOGITS
     surla
    0.41
     Вікіпе
    0.33
     fortæ
    0.33
    extAlignment
    0.32
     rhestr
    0.28
    SneakyThrows
    0.27
     filós
    0.26
     désolés
    0.26
    Jegyzetek
    0.26
    ClientRect
    0.25
    Activations Density 36.179%

    No Known Activations