© 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. 31-TRANSCODER-HP
    7. 18380
    Prev
    Next
    INDEX
    Explanations

    Code and technical discussions

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_31.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.31.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ques
    -0.31
    asy
    -0.29
    arious
    -0.28
    heim
    -0.27
     TEST
    -0.27
    (TEST
    -0.26
    inous
    -0.26
    gress
    -0.25
    mars
    -0.25
    _TestCase
    -0.25
    POSITIVE LOGITS
    åĩ¿
    0.24
     Watches
    0.23
    ມ
    0.23
    usu
    0.23
    Rot
    0.23
    ToProps
    0.23
    æĨĭ
    0.23
     exhib
    0.23
     findById
    0.23
     اÙĦبÙĦ
    0.23
    Activations Density 0.011%

    No Known Activations