© 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. 18620
    Prev
    Next
    INDEX
    Explanations

    chick

    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
    æ¸ł
    -0.27
    ä½įåĪĹ
    -0.27
    æİĴè¡Į
    -0.26
    олов
    -0.25
     misrepresented
    -0.25
    morgan
    -0.25
    lander
    -0.25
    å¾Ĺå¾Ī
    -0.25
     Huck
    -0.25
     liberty
    -0.25
    POSITIVE LOGITS
    ens
    0.29
     staying
    0.28
    ember
    0.28
    佩
    0.26
    een
    0.25
    (enabled
    0.25
     stayed
    0.25
    åī¥
    0.24
    -fe
    0.24
     fian
    0.24
    Activations Density 0.058%

    No Known Activations