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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 131k
    9. 20-GEMMASCOPE-RES-131K
    11. 4512
    Prev
    Next
    INDEX
    Explanations

    declarations and instructions related to health, personal conduct, and customization

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

    instead, rather, other options

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

    Detects directives and instructions—imperative or advisory phrasing that tells someone to do or not do something (e.g., usage/dosage and moderation guidance).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
    BeginInit
    -0.71
     propOrder
    -0.69
    ineno
    -0.54
    ErrUnexpectedEOF
    -0.51
    ühungen
    -0.49
    
    -0.47
     surla
    -0.47
    AddAttribute
    -0.46
    PerformLayout
    -0.46
    ьаж
    -0.45
    POSITIVE LOGITS
     instead
    0.45
     invece
    0.43
     Instead
    0.36
    Instead
    0.35
    WebElementEntity
    0.35
     rather
    0.34
     vielmehr
    0.34
    tool
    0.34
    instead
    0.33
    $_['
    0.32
    Activations Density 0.095%

    No Known Activations