© 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. 18-TRANSCODER-HP
    7. 88931
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_18.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.18.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    onian
    -0.28
    èĦī
    -0.28
    deo
    -0.27
    èĦĪ
    -0.26
     pos
    -0.25
     Rei
    -0.24
     giả
    -0.23
    ert
    -0.23
    etections
    -0.23
    igo
    -0.23
    POSITIVE LOGITS
     timers
    0.26
    è·ŁæĪij说
    0.26
    [".
    0.26
    ologia
    0.25
    #Region
    0.25
    é©Ń
    0.25
    ÑĥÑĩа
    0.24
    å·Ŀ
    0.24
     comprom
    0.23
    éļĨ
    0.23
    Activations Density 0.925%

    No Known Activations

    This feature has no known activations.