© 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. 19160
    Prev
    Next
    INDEX
    Explanations

    legislature/congress

    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
    åĩ»
    -0.32
    irma
    -0.26
    jah
    -0.25
     gl
    -0.25
     fried
    -0.25
    åĶIJæľĿ
    -0.24
    ä¸Ńæĸ°ç½ij
    -0.24
     TK
    -0.23
     LV
    -0.23
     compl
    -0.23
    POSITIVE LOGITS
    æݧåĪ¶
    0.33
     control
    0.32
    -controlled
    0.31
     CONTROL
    0.31
    control
    0.29
     controlled
    0.28
     intent
    0.28
    (control
    0.27
    contro
    0.27
    District
    0.27
    Activations Density 0.981%

    No Known Activations