INDEX
    Explanations

    code or documentation

    The neuron fires on bits of document‐level markup or boilerplate—things like HTML tags, man‐page section markers, code‐formatting delimiters, and other structural markup rather than ordinary prose.

    New Auto-Interp
    Negative Logits
    .d
    -0.06
    Cart
    -0.06
    egas
    -0.06
    Approved
    -0.06
    θήκη
    -0.06
    »
    -0.06
    .details
    -0.06
    oca
    -0.06
     asshole
    -0.06
    estar
    -0.06
    POSITIVE LOGITS
    (utf
    0.06
     Ej
    0.06
     včetně
    0.06
    PointF
    0.06
     desde
    0.06
     Come
    0.06
    ользов
    0.06
    ाइट
    0.06
     Styles
    0.06
     जम
    0.06
    Act Density 0.005%

    No Known Activations