© 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. 35-TRANSCODER-HP
    7. 39469
    Prev
    Next
    INDEX
    Explanations

    Abbreviations/symbols in documents

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_35.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.35.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    è´Łè´£
    -0.27
     clearly
    -0.26
     compet
    -0.25
    cil
    -0.25
    cls
    -0.25
    ä¸įèĪį
    -0.25
     kcal
    -0.24
     nicely
    -0.24
    ä¸įåı¯
    -0.24
    uvre
    -0.24
    POSITIVE LOGITS
    atti
    0.30
    ting
    0.28
    atif
    0.27
    èĨ¨
    0.26
    atten
    0.26
    鼷éľĨ
    0.26
    ITES
    0.26
    nt
    0.26
    ATING
    0.25
     Transitional
    0.25
    Activations Density 0.045%

    No Known Activations