© 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
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 2795
    Prev
    Next
    INDEX
    Explanations

    URLs and references to news articles or stories

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

    URLs and web references, particularly from the Mirror.co.uk news site.

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi

    markers of news articles and web links—URL structure, headline-style capitalization/punctuation (slashes, hyphens, apostrophes), and prominent named entities.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    line breaks or paragraph boundaries in text, particularly in news articles or web content.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    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
    Tikang
    -0.78
    AndEndTag
    -0.76
    yarnpkg
    -0.71
    ReusableCell
    -0.71
     CreateTagHelper
    -0.70
    RenderAtEndOf
    -0.69
     ModelExpression
    -0.68
    GOTREF
    -0.66
    OGND
    -0.65
     erſt
    -0.64
    POSITIVE LOGITS
     føl
    0.28
     deal
    0.27
    deal
    0.26
     flexibilidad
    0.25
    ά
    0.25
    News
    0.25
    '
    0.23
     polisi
    0.23
    alpha
    0.23
    PostInfinity
    0.23
    Activations Density 0.072%

    No Known Activations