© 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. 33-TRANSCODER-HP
    7. 54032
    Prev
    Next
    INDEX
    Explanations

    Letter combinations

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_33.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.33.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    æ·±åĬłå·¥
    -0.26
    è°ĭ
    -0.26
    bidden
    -0.24
    ewire
    -0.24
    -rays
    -0.24
    tees
    -0.23
    omination
    -0.23
    空ä¸Ń
    -0.23
     compositions
    -0.23
    ompson
    -0.23
    POSITIVE LOGITS
    -L
    0.38
    -Jul
    0.38
    -D
    0.36
    -R
    0.34
    -V
    0.34
    -Bar
    0.33
    -M
    0.32
    -C
    0.32
    -P
    0.32
    -A
    0.31
    Activations Density 0.053%

    No Known Activations