© 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. 30-TRANSCODER-HP
    7. 96461
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_30.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.30.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    deer
    -0.30
    quo
    -0.27
    ours
    -0.26
     wik
    -0.26
    leases
    -0.25
    lectron
    -0.24
    Pocket
    -0.24
    ucs
    -0.24
    oard
    -0.24
    Blend
    -0.23
    POSITIVE LOGITS
    otics
    0.28
    èĸ¯
    0.25
     Principle
    0.24
     revealing
    0.24
    ález
    0.24
     nearer
    0.23
     principle
    0.23
     disclose
    0.23
     maid
    0.23
     Government
    0.23
    Activations Density 0.101%

    No Known Activations

    This feature has no known activations.