© 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. 18147
    Prev
    Next
    INDEX
    Explanations

    Code and function names

    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
    ined
    -0.30
     finely
    -0.28
    INED
    -0.28
    âľķ
    -0.27
    ней
    -0.25
    æĥ©æĪĴ
    -0.25
     finalize
    -0.25
    feit
    -0.24
    riger
    -0.24
    à¸Ħว
    -0.24
    POSITIVE LOGITS
    enga
    0.31
    dba
    0.28
    enterprise
    0.28
    CAD
    0.27
     Kata
    0.26
     CHO
    0.25
    UGC
    0.25
     choice
    0.25
    picked
    0.25
    avana
    0.25
    Activations Density 0.105%

    No Known Activations