© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26
    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    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. 3887
    Prev
    Next
    INDEX
    Explanations

    the start of sections or paragraphs in documents

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot

    numeric sequences and math expressions

    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
    Negative Logits
    webElementXpaths
    -1.05
    MemoryWarning
    -1.00
     CreateTagHelper
    -0.99
     }}$}
    -0.94
    Билгалдахарш
    -0.92
     Efq
    -0.88
    .")]
    -0.87
    ."));
    -0.85
    .}~\
    -0.84
    .)}
    -0.84
    POSITIVE LOGITS
     However
    1.46
     Therefore
    1.29
    However
    1.25
     This
    1.25
     Unfortunately
    1.22
     Moreover
    1.22
     Then
    1.18
     Furthermore
    1.17
     Thus
    1.16
     Although
    1.15
    Activations Density 0.987%

    No Known Activations