© 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. 98589
    Prev
    Next
    INDEX
    Explanations

    secret

    np_max-act · gemini-2.0-flash
    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
     occupying
    -0.29
    ded
    -0.26
    bang
    -0.25
    esk
    -0.25
    åįłæį®
    -0.24
    ãģĹãģĭãģªãģĦ
    -0.24
    親
    -0.24
    ,X
    -0.24
    dg
    -0.24
     Only
    -0.24
    POSITIVE LOGITS
    lain
    0.36
    ç§ĺ
    0.30
    ç§ĺå¯Ĩ
    0.29
    éļIJèĹı
    0.29
     hidden
    0.29
     implic
    0.28
     hide
    0.28
    hatt
    0.27
     secretly
    0.27
     szer
    0.27
    Activations Density 0.054%

    No Known Activations