Update spaCy pipeline
cf79cce
|
{ |
|
"punct_chars":[ |
|
"\u1c3b", |
|
"\u166e", |
|
"\ud805\ude41", |
|
"\u1c7e", |
|
"\u1736", |
|
"\ud836\ude88", |
|
"\uaa5f", |
|
"\u1b5f", |
|
"\ud81a\ude6f", |
|
"\ud805\uddd7", |
|
"\u0589", |
|
"\ud804\uddc5", |
|
"\ud805\udc4c", |
|
"\u1803", |
|
"\ud804\udcc1", |
|
"\ud805\udf3e", |
|
"\ua4ff", |
|
"\u1aa8", |
|
"\ud807\udc42", |
|
"\ud81a\udf38", |
|
"\ua8ce", |
|
"\ud804\udd42", |
|
"\u1b5b", |
|
"\ud804\udddf", |
|
"\ud806\ude9c", |
|
"\ud805\uddc2", |
|
"\ufe56", |
|
"\ud805\uddc3", |
|
"\ud804\uddc6", |
|
"\uff61", |
|
"\ud805\uddcd", |
|
"\ua876", |
|
"\uaa5e", |
|
"\u1944", |
|
"\uff01", |
|
"\ufe52", |
|
"\ud802\ude56", |
|
"\ud805\uddd5", |
|
"\u104a", |
|
"\u1c3c", |
|
"\ud805\uddc9", |
|
"\uabeb", |
|
"\ud804\udcbe", |
|
"\u1aaa", |
|
"\ud802\ude57", |
|
"\ud804\ude3c", |
|
"\ua877", |
|
"\ud82f\udc9f", |
|
"\ua9c9", |
|
"\ud804\udc48", |
|
"\ud804\ude3b", |
|
"\u0701", |
|
"\u2049", |
|
"\uaaf1", |
|
"\ud804\uddde", |
|
"\u3002", |
|
"\u2e3c", |
|
"\ud804\udcbf", |
|
"\u2047", |
|
"\ud805\uddd1", |
|
"\uaaf0", |
|
"\ud806\ude42", |
|
"\u1b5e", |
|
"\u1b5a", |
|
"\u0964", |
|
"\u1c7f", |
|
"\ud805\udc4b", |
|
"\ua9c8", |
|
"\ud805\uddcf", |
|
"\ud804\ude39", |
|
"\ud805\uddcc", |
|
"\ua8cf", |
|
"\ud805\uddd3", |
|
"\ud806\ude9b", |
|
"\ud81a\ude6e", |
|
"\ud804\udcc0", |
|
"\ud805\uddca", |
|
"\u2048", |
|
"\ud804\udc47", |
|
"\u1aab", |
|
"\ud804\udd43", |
|
"\ud807\udc41", |
|
"!", |
|
"\uff1f", |
|
"\u1809", |
|
"\ud805\udf3c", |
|
"\ud805\uddd2", |
|
"\ua6f7", |
|
"\u06d4", |
|
"\ud804\udea9", |
|
"\ud81a\udf44", |
|
"\ua60e", |
|
"\u061f", |
|
"\u104b", |
|
"\u1368", |
|
"\u2e2e", |
|
"\ua6f3", |
|
"?", |
|
"\u1945", |
|
"\ufe57", |
|
"\ud81a\udf37", |
|
"\u0965", |
|
"\ud81a\udef5", |
|
"\u203d", |
|
"\ud805\uddce", |
|
"\ud805\uddd6", |
|
".", |
|
"\ud804\uddcd", |
|
"\u0702", |
|
"\ud805\udf3d", |
|
"\ua60f", |
|
"\ud805\uddd4", |
|
"\ua92f", |
|
"\ud804\ude38", |
|
"\uaa5d", |
|
"\ud805\uddcb", |
|
"\ud806\ude43", |
|
"\u0700", |
|
"\u1367", |
|
"\u1362", |
|
"\ud804\udd41", |
|
"\ud805\ude42", |
|
"\u1aa9", |
|
"\u07f9", |
|
"\ud805\uddd0", |
|
"\u1735", |
|
"\uff0e", |
|
"\u203c" |
|
], |
|
"overwrite":false |
|
} |
