© 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. Qwen3-4B
    5. 20-TRANSCODER-HP
    7. 107074
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_20.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.20.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    []{
    -0.28
    []{↵
    -0.28
     pads
    -0.26
    oru
    -0.26
    cue
    -0.25
    iac
    -0.25
     PANEL
    -0.25
     Cue
    -0.25
    ç«ĻçĿĢ
    -0.24
     système
    -0.24
    POSITIVE LOGITS
    æĪIJç«ĭ
    0.25
    èµĽ
    0.25
    楼æĪ¿
    0.25
    oba
    0.25
    gett
    0.25
    atter
    0.24
    é«ĺ峰论åĿĽ
    0.24
    è·ŁçĿĢ
    0.24
     ther
    0.24
    丼
    0.24
    Activations Density 0.477%

    No Known Activations

    This feature has no known activations.