INDEX
    Explanations

    The neuron consistently activates on HTML markup tokens (e.g. tag names and delimiters like `<div>`, `<h2>`, `</div>`, attribute names like `text/javascript`, etc.), i.e. it’s looking for HTML tags/markup.

    New Auto-Interp
    Negative Logits
    fest
    -0.07
     Hands
    -0.07
     bedroom
    -0.07
     beams
    -0.07
    日期
    -0.07
    รายงาน
    -0.07
    particle
    -0.07
     Jerseys
    -0.06
     каждого
    -0.06
    433
    -0.06
    POSITIVE LOGITS
     Mehmet
    0.06
    شنبه
    0.06
    Choose
    0.06
     čin
    0.06
    contents
    0.06
    ransition
    0.06
    )prepare
    0.06
    asionally
    0.06
    (mid
    0.06
    unordered
    0.06
    Act Density 0.015%

    No Known Activations