© 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. 19351
    Prev
    Next
    INDEX
    Explanations

    on

    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
     "|"
    -0.29
    éĥ½éľĢè¦ģ
    -0.27
    ä¸įåıªæĺ¯
    -0.27
    rö
    -0.25
     Need
    -0.25
    breaker
    -0.25
    ACK
    -0.25
    (range
    -0.24
    YC
    -0.24
    çļĦä¸Ģéĥ¨åĪĨ
    -0.24
    POSITIVE LOGITS
     external
    0.43
     having
    0.38
     maintaining
    0.36
     being
    0.35
     managing
    0.34
     their
    0.34
     storing
    0.34
     capturing
    0.34
     its
    0.34
     consuming
    0.33
    Activations Density 0.148%

    No Known Activations