© 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. 129478
    Prev
    Next
    INDEX
    Explanations

    Period

    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
    kea
    -0.28
    ongsTo
    -0.28
    帮ä»ĸ
    -0.26
    uÅŁ
    -0.25
    诱人
    -0.25
    upy
    -0.25
    gere
    -0.25
    kart
    -0.25
    umpt
    -0.24
    aku
    -0.24
    POSITIVE LOGITS
    æµĨ
    0.27
    ation
    0.26
     thinks
    0.25
    /comment
    0.25
     think
    0.25
    رÙĬÙģ
    0.24
    å¦Ĥæľīä¾µæĿĥ
    0.24
    ucky
    0.24
     Expert
    0.24
    expert
    0.24
    Activations Density 9.881%

    No Known Activations