© 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. 33-TRANSCODER-HP
    7. 52952
    Prev
    Next
    INDEX
    Explanations

    say "s ending"

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_33.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.33.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    .uf
    -0.25
     hol
    -0.25
     Seymour
    -0.25
    _fun
    -0.23
    SOURCE
    -0.23
    hci
    -0.23
    gba
    -0.23
    ToggleButton
    -0.23
    RTOS
    -0.23
    blink
    -0.22
    POSITIVE LOGITS
    ences
    0.39
    oids
    0.35
    es
    0.30
    s
    0.30
    elines
    0.28
    tons
    0.27
    vars
    0.27
    isses
    0.27
    елÑĮ
    0.26
    tion
    0.26
    Activations Density 0.008%

    No Known Activations