© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-4B
    5. 33-TRANSCODER-HP
    7. 54578
    Prev
    Next
    INDEX
    Explanations

    of

    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
    Negative Logits
     qu
    -0.27
     den
    -0.26
    è¹ij
    -0.26
    éĴİ
    -0.25
    _mc
    -0.25
    fdc
    -0.25
    fs
    -0.25
    cac
    -0.25
     forward
    -0.25
    fsp
    -0.24
    POSITIVE LOGITS
    -Life
    0.33
     Light
    0.31
     Difficulty
    0.31
     Tea
    0.28
     Knowledge
    0.28
     Interest
    0.27
     Time
    0.27
    ä¸Ģæīĭ
    0.27
     Birth
    0.27
     Life
    0.27
    Activations Density 0.013%

    No Known Activations