© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-4B
    5. 31-TRANSCODER-HP
    7. 19809
    Prev
    Next
    INDEX
    Explanations

    personal experiences

    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
    Negative Logits
     surprise
    -0.46
     surprises
    -0.45
     unexpected
    -0.45
     surprising
    -0.42
    æĦıå¤ĸ
    -0.37
    å¥ĩ
    -0.35
    æĦıæĥ³ä¸įåĪ°
    -0.34
     Unexpected
    -0.34
     strang
    -0.33
     puzz
    -0.31
    POSITIVE LOGITS
     flo
    0.33
     taken
    0.28
    asted
    0.28
    -dat
    0.26
    .IsNull
    0.26
     murdered
    0.26
    赶æĿ¥
    0.25
     dol
    0.25
    IsNull
    0.24
     absolutely
    0.24
    Activations Density 0.260%

    No Known Activations