© 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. Gemma-2-2B
    5. 20-GEMMASCOPE-TRANSCODER-16K
    7. 13256
    Prev
    Next
    INDEX
    Explanations

    the word "refer" and its inflections, sometimes with a weak activation for surrounding words

    oai_token-act-pair · gemini-2.0-flash

    refer

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_20/width_16k/average_l0_11
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.20.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     involves
    -1.19
     involve
    -1.19
     refers
    -1.11
     revolves
    -1.07
     pertains
    -1.06
     relates
    -1.05
     revolve
    -1.05
     consiste
    -1.05
     consists
    -1.03
     consist
    -0.99
    POSITIVE LOGITS
    īdz
    0.38
    ydın
    0.37
    CopyWith
    0.37
     ostavi
    0.35
    ระบบ
    0.34
     Exacts
    0.34
    ?>">
    0.34
    iden
    0.34
    iiiiiiii
    0.33
    主持
    0.33
    Activations Density 1.392%

    No Known Activations