INDEX
    Explanations

    program names and trademark symbols

    the neuron detects brand/product/software names and other proper nouns (named entities) often appearing in recommendation or commerce contexts.

    New Auto-Interp
    Negative Logits
    و
    0.21
     Besuch
    0.20
    حده
    0.20
    人家
    0.20
    ع
    0.20
    لا
    0.20
    ل
    0.19
    elesen
    0.19
     Gegner
    0.18
    _
    0.18
    POSITIVE LOGITS
    ™.
    0.36
    ™,
    0.34
    0.33
    ®.
    0.32
    ®,
    0.30
    ®
    0.24
     combines
    0.23
    ᴿ
    0.22
    0.22
    0.22
    Act Density 0.440%

    No Known Activations