© 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. 20-TRANSCODER-HP
    7. 129595
    Prev
    Next
    INDEX
    Explanations

    already existing

    np_max-act · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_20.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.20.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Descriptors
    -0.26
     Accord
    -0.25
     Cave
    -0.25
     Alignment
    -0.24
     align
    -0.24
     Lig
    -0.23
    atories
    -0.23
    liches
    -0.23
     Accepted
    -0.23
    ĥģ
    -0.23
    POSITIVE LOGITS
    约åIJĪ
    0.29
    å¢ĹãģĪ
    0.27
     zusammen
    0.26
    èĪŀ
    0.25
    رÙĬÙĤ
    0.25
    éĥ½è¢«
    0.24
    ivery
    0.24
    degrees
    0.24
    STEP
    0.24
     ruining
    0.24
    Activations Density 9.029%

    No Known Activations