© 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. 39128
    Prev
    Next
    INDEX
    Explanations

    say "ed ending"

    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
    ie
    -0.29
     can
    -0.29
     do
    -0.28
     spiral
    -0.28
     and
    -0.28
     in
    -0.28
    case
    -0.28
     or
    -0.27
     be
    -0.27
     is
    -0.27
    POSITIVE LOGITS
    uted
    0.39
    arded
    0.38
    åIJ¬è¿ĩ
    0.37
    enced
    0.37
    ened
    0.36
    ified
    0.36
     mastered
    0.36
    ited
    0.36
    atched
    0.36
    eded
    0.35
    Activations Density 65.767%

    No Known Activations