© 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. Gemma-2-2B
    5. 2-GEMMASCOPE-TRANSCODER-16K
    7. 10490
    Prev
    Next
    INDEX
    Explanations

    German language modal verbs and auxiliary verbs, specifically in the context of complex sentence structures.

    oai_token-act-pair · gemini-2.0-flash

    German "werden"

    np_max-act · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_2/width_16k/average_l0_49
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.2.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
     werden
    -1.09
     wird
    -1.07
     wurde
    -0.93
     werd
    -0.81
     wurden
    -0.81
     können
    -0.76
     kann
    -0.73
    Wird
    -0.68
    wird
    -0.66
     worden
    -0.63
    POSITIVE LOGITS
     AssemblyCulture
    1.07
    :✨
    1.05
    MLLoader
    1.04
     '\\;'
    1.01
     kasarigan
    1.01
    AddTagHelper
    0.98
    aarrggbb
    0.92
    SourceChecksum
    0.91
    poptosis
    0.90
     Normdatei
    0.90
    Activations Density 0.745%

    No Known Activations