© 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. 53485
    Prev
    Next
    INDEX
    Explanations

    code or file paths

    np_max-act-logits · gemini-2.0-flash
    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
    åĬ³
    -0.29
    eko
    -0.28
     pyt
    -0.27
    游
    -0.26
    (head
    -0.26
    ibold
    -0.26
    eger
    -0.25
     SALE
    -0.25
    åĩłä½ķ
    -0.24
    ean
    -0.24
    POSITIVE LOGITS
    ulous
    0.29
    "She
    0.26
    åķģ
    0.25
     Wenger
    0.25
    ãģŀ
    0.25
    ENCH
    0.24
    naz
    0.24
     turf
    0.24
     rect
    0.24
    å¸Ļ
    0.24
    Activations Density 0.001%

    No Known Activations