© 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. 129833
    Prev
    Next
    INDEX
    Explanations

    Code and error messages

    np_max-act · gemini-2.0-flash
    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
    že
    -0.28
    GBT
    -0.27
    isans
    -0.26
     thirsty
    -0.26
    inci
    -0.26
    oric
    -0.25
     sane
    -0.25
    ixe
    -0.25
     drinkers
    -0.25
    -Cs
    -0.24
    POSITIVE LOGITS
    Enter
    0.28
    ktop
    0.27
     proc
    0.27
    Land
    0.26
    uzu
    0.26
    _Anim
    0.25
    VIRTUAL
    0.24
     Proc
    0.24
     meta
    0.24
    land
    0.24
    Activations Density 7.672%

    No Known Activations