© 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. 53709
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    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
    velle
    -0.27
     priorities
    -0.26
     maps
    -0.26
    scripts
    -0.25
    meld
    -0.24
    åªĴ
    -0.24
     glGet
    -0.24
    æ®ĸ
    -0.24
    rides
    -0.24
     marc
    -0.24
    POSITIVE LOGITS
    FOUNDATION
    0.29
     arising
    0.29
    pytest
    0.27
    æľįåĬ¡å¹³åı°
    0.26
    壮
    0.25
    ä¼ĹæīĢ
    0.25
    ar
    0.25
    lik
    0.24
    atform
    0.24
    aton
    0.24
    Activations Density 0.019%

    No Known Activations

    This feature has no known activations.