© 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. 22-TRANSCODER-HP
    7. 111755
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_22.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.22.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    FACE
    -0.29
     FACE
    -0.27
    uche
    -0.25
    Tabs
    -0.23
    ell
    -0.23
    (reply
    -0.23
     ByteArrayInputStream
    -0.23
     tinder
    -0.23
     Anc
    -0.23
    createClass
    -0.23
    POSITIVE LOGITS
     screen
    0.27
    /screen
    0.27
     miss
    0.26
    涯
    0.25
    unks
    0.25
    misión
    0.24
    imiter
    0.24
    pletion
    0.24
    ovenant
    0.24
    screen
    0.24
    Activations Density 0.001%

    No Known Activations

    This feature has no known activations.