Upload tokenizer
Browse files- vocab.json +100 -100
vocab.json
CHANGED
|
@@ -1,104 +1,104 @@
|
|
| 1 |
{
|
| 2 |
-
"'":
|
| 3 |
-
"R":
|
| 4 |
-
"U":
|
| 5 |
-
"[":
|
| 6 |
"[PAD]": 101,
|
| 7 |
"[UNK]": 100,
|
| 8 |
-
"]":
|
| 9 |
-
"_":
|
| 10 |
-
"`":
|
| 11 |
-
"a":
|
| 12 |
-
"b":
|
| 13 |
-
"d":
|
| 14 |
-
"e":
|
| 15 |
-
"f":
|
| 16 |
-
"h":
|
| 17 |
-
"i":
|
| 18 |
-
"k":
|
| 19 |
-
"m":
|
| 20 |
-
"r":
|
| 21 |
-
"t":
|
| 22 |
-
"u":
|
| 23 |
-
"w":
|
| 24 |
-
"|":
|
| 25 |
-
"،":
|
| 26 |
-
"ؔ":
|
| 27 |
-
"؛":
|
| 28 |
-
"؟":
|
| 29 |
-
"ء":
|
| 30 |
-
"آ":
|
| 31 |
-
"أ":
|
| 32 |
-
"ؤ":
|
| 33 |
-
"ئ":
|
| 34 |
-
"ا":
|
| 35 |
-
"ب":
|
| 36 |
-
"ت":
|
| 37 |
-
"ث":
|
| 38 |
-
"ج":
|
| 39 |
-
"ح":
|
| 40 |
-
"خ":
|
| 41 |
-
"د":
|
| 42 |
-
"ذ":
|
| 43 |
-
"ر":
|
| 44 |
-
"ز":
|
| 45 |
-
"س":
|
| 46 |
-
"ش":
|
| 47 |
-
"ص":
|
| 48 |
-
"ض":
|
| 49 |
-
"ط":
|
| 50 |
-
"ظ":
|
| 51 |
-
"ع":
|
| 52 |
-
"غ":
|
| 53 |
-
"ف":
|
| 54 |
-
"ق":
|
| 55 |
-
"ل":
|
| 56 |
-
"م":
|
| 57 |
-
"ن":
|
| 58 |
-
"ه":
|
| 59 |
-
"و":
|
| 60 |
-
"ي":
|
| 61 |
-
"ً":
|
| 62 |
-
"َ":
|
| 63 |
-
"ُ":
|
| 64 |
-
"ِ":
|
| 65 |
-
"ّ":
|
| 66 |
-
"ْ":
|
| 67 |
-
"ٓ":
|
| 68 |
-
"ٔ":
|
| 69 |
-
"ٗ":
|
| 70 |
-
"ٰ":
|
| 71 |
-
"ٴ":
|
| 72 |
-
"ٹ":
|
| 73 |
-
"پ":
|
| 74 |
-
"چ":
|
| 75 |
-
"ڈ":
|
| 76 |
-
"ڑ":
|
| 77 |
-
"ژ":
|
| 78 |
-
"ک":
|
| 79 |
-
"گ":
|
| 80 |
-
"ں":
|
| 81 |
-
"ھ":
|
| 82 |
-
"ہ":
|
| 83 |
-
"ۂ":
|
| 84 |
-
"ۃ":
|
| 85 |
-
"ی":
|
| 86 |
-
"ے":
|
| 87 |
-
"ۓ":
|
| 88 |
-
"۔":
|
| 89 |
-
"‘":
|
| 90 |
-
"’":
|
| 91 |
-
"ﭨ":
|
| 92 |
-
"ﮭ":
|
| 93 |
-
"ﮯ":
|
| 94 |
-
"ﯾ":
|
| 95 |
-
"ﷲ":
|
| 96 |
-
"ﷺ":
|
| 97 |
-
"ﺅ":
|
| 98 |
-
"ﺗ":
|
| 99 |
-
"ﺘ":
|
| 100 |
-
"ﺩ":
|
| 101 |
-
"ﺲ":
|
| 102 |
-
"ﻧ":
|
| 103 |
-
"ﻮ":
|
| 104 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"'": 60,
|
| 3 |
+
"R": 55,
|
| 4 |
+
"U": 10,
|
| 5 |
+
"[": 80,
|
| 6 |
"[PAD]": 101,
|
| 7 |
"[UNK]": 100,
|
| 8 |
+
"]": 88,
|
| 9 |
+
"_": 68,
|
| 10 |
+
"`": 76,
|
| 11 |
+
"a": 40,
|
| 12 |
+
"b": 49,
|
| 13 |
+
"d": 78,
|
| 14 |
+
"e": 69,
|
| 15 |
+
"f": 20,
|
| 16 |
+
"h": 73,
|
| 17 |
+
"i": 97,
|
| 18 |
+
"k": 8,
|
| 19 |
+
"m": 71,
|
| 20 |
+
"r": 7,
|
| 21 |
+
"t": 14,
|
| 22 |
+
"u": 65,
|
| 23 |
+
"w": 2,
|
| 24 |
+
"|": 67,
|
| 25 |
+
"،": 87,
|
| 26 |
+
"ؔ": 75,
|
| 27 |
+
"؛": 47,
|
| 28 |
+
"؟": 61,
|
| 29 |
+
"ء": 12,
|
| 30 |
+
"آ": 98,
|
| 31 |
+
"أ": 39,
|
| 32 |
+
"ؤ": 21,
|
| 33 |
+
"ئ": 46,
|
| 34 |
+
"ا": 66,
|
| 35 |
+
"ب": 9,
|
| 36 |
+
"ت": 59,
|
| 37 |
+
"ث": 99,
|
| 38 |
+
"ج": 42,
|
| 39 |
+
"ح": 81,
|
| 40 |
+
"خ": 25,
|
| 41 |
+
"د": 33,
|
| 42 |
+
"ذ": 26,
|
| 43 |
+
"ر": 36,
|
| 44 |
+
"ز": 58,
|
| 45 |
+
"س": 86,
|
| 46 |
+
"ش": 32,
|
| 47 |
+
"ص": 11,
|
| 48 |
+
"ض": 17,
|
| 49 |
+
"ط": 24,
|
| 50 |
+
"ظ": 3,
|
| 51 |
+
"ع": 83,
|
| 52 |
+
"غ": 15,
|
| 53 |
+
"ف": 56,
|
| 54 |
+
"ق": 19,
|
| 55 |
+
"ل": 41,
|
| 56 |
+
"م": 57,
|
| 57 |
+
"ن": 27,
|
| 58 |
+
"ه": 5,
|
| 59 |
+
"و": 70,
|
| 60 |
+
"ي": 74,
|
| 61 |
+
"ً": 50,
|
| 62 |
+
"َ": 4,
|
| 63 |
+
"ُ": 44,
|
| 64 |
+
"ِ": 1,
|
| 65 |
+
"ّ": 45,
|
| 66 |
+
"ْ": 23,
|
| 67 |
+
"ٓ": 0,
|
| 68 |
+
"ٔ": 77,
|
| 69 |
+
"ٗ": 29,
|
| 70 |
+
"ٰ": 95,
|
| 71 |
+
"ٴ": 16,
|
| 72 |
+
"ٹ": 37,
|
| 73 |
+
"پ": 22,
|
| 74 |
+
"چ": 91,
|
| 75 |
+
"ڈ": 90,
|
| 76 |
+
"ڑ": 84,
|
| 77 |
+
"ژ": 48,
|
| 78 |
+
"ک": 89,
|
| 79 |
+
"گ": 34,
|
| 80 |
+
"ں": 35,
|
| 81 |
+
"ھ": 53,
|
| 82 |
+
"ہ": 13,
|
| 83 |
+
"ۂ": 51,
|
| 84 |
+
"ۃ": 82,
|
| 85 |
+
"ی": 62,
|
| 86 |
+
"ے": 94,
|
| 87 |
+
"ۓ": 30,
|
| 88 |
+
"۔": 18,
|
| 89 |
+
"‘": 96,
|
| 90 |
+
"’": 92,
|
| 91 |
+
"ﭨ": 6,
|
| 92 |
+
"ﮭ": 72,
|
| 93 |
+
"ﮯ": 93,
|
| 94 |
+
"ﯾ": 31,
|
| 95 |
+
"ﷲ": 79,
|
| 96 |
+
"ﷺ": 38,
|
| 97 |
+
"ﺅ": 54,
|
| 98 |
+
"ﺗ": 64,
|
| 99 |
+
"ﺘ": 85,
|
| 100 |
+
"ﺩ": 63,
|
| 101 |
+
"ﺲ": 52,
|
| 102 |
+
"ﻧ": 43,
|
| 103 |
+
"ﻮ": 28
|
| 104 |
}
|