Upload tokenizer
Browse files- vocab.json +43 -43
vocab.json
CHANGED
|
@@ -1,51 +1,51 @@
|
|
| 1 |
{
|
| 2 |
"[PAD]": 48,
|
| 3 |
"[UNK]": 47,
|
| 4 |
-
"c":
|
| 5 |
-
"i":
|
| 6 |
-
"o":
|
| 7 |
-
"y":
|
| 8 |
-
"|":
|
| 9 |
-
"«":
|
| 10 |
-
"»":
|
| 11 |
-
"ç":
|
| 12 |
-
"а":
|
| 13 |
"б": 0,
|
| 14 |
-
"в":
|
| 15 |
-
"г":
|
| 16 |
-
"д":
|
| 17 |
-
"е":
|
| 18 |
-
"ж":
|
| 19 |
-
"з":
|
| 20 |
-
"и":
|
| 21 |
-
"й":
|
| 22 |
-
"к":
|
| 23 |
"л": 35,
|
| 24 |
-
"м":
|
| 25 |
-
"н":
|
| 26 |
-
"о":
|
| 27 |
-
"п":
|
| 28 |
-
"р":
|
| 29 |
-
"с":
|
| 30 |
-
"т":
|
| 31 |
-
"у":
|
| 32 |
-
"ф":
|
| 33 |
"х": 32,
|
| 34 |
-
"ц":
|
| 35 |
-
"ч":
|
| 36 |
-
"ш":
|
| 37 |
-
"щ":
|
| 38 |
-
"ъ":
|
| 39 |
"ы": 45,
|
| 40 |
-
"ь":
|
| 41 |
-
"э":
|
| 42 |
-
"ю":
|
| 43 |
-
"я":
|
| 44 |
-
"і":
|
| 45 |
-
"ј":
|
| 46 |
-
"ҕ":
|
| 47 |
-
"ҥ":
|
| 48 |
-
"ү":
|
| 49 |
-
"һ":
|
| 50 |
-
"ө":
|
| 51 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"[PAD]": 48,
|
| 3 |
"[UNK]": 47,
|
| 4 |
+
"c": 24,
|
| 5 |
+
"i": 15,
|
| 6 |
+
"o": 30,
|
| 7 |
+
"y": 6,
|
| 8 |
+
"|": 23,
|
| 9 |
+
"«": 16,
|
| 10 |
+
"»": 19,
|
| 11 |
+
"ç": 22,
|
| 12 |
+
"а": 31,
|
| 13 |
"б": 0,
|
| 14 |
+
"в": 13,
|
| 15 |
+
"г": 12,
|
| 16 |
+
"д": 17,
|
| 17 |
+
"е": 38,
|
| 18 |
+
"ж": 14,
|
| 19 |
+
"з": 8,
|
| 20 |
+
"и": 28,
|
| 21 |
+
"й": 42,
|
| 22 |
+
"к": 27,
|
| 23 |
"л": 35,
|
| 24 |
+
"м": 20,
|
| 25 |
+
"н": 25,
|
| 26 |
+
"о": 10,
|
| 27 |
+
"п": 33,
|
| 28 |
+
"р": 29,
|
| 29 |
+
"с": 26,
|
| 30 |
+
"т": 11,
|
| 31 |
+
"у": 18,
|
| 32 |
+
"ф": 3,
|
| 33 |
"х": 32,
|
| 34 |
+
"ц": 39,
|
| 35 |
+
"ч": 40,
|
| 36 |
+
"ш": 4,
|
| 37 |
+
"щ": 21,
|
| 38 |
+
"ъ": 9,
|
| 39 |
"ы": 45,
|
| 40 |
+
"ь": 34,
|
| 41 |
+
"э": 1,
|
| 42 |
+
"ю": 5,
|
| 43 |
+
"я": 36,
|
| 44 |
+
"і": 46,
|
| 45 |
+
"ј": 37,
|
| 46 |
+
"ҕ": 44,
|
| 47 |
+
"ҥ": 43,
|
| 48 |
+
"ү": 41,
|
| 49 |
+
"һ": 7,
|
| 50 |
+
"ө": 2
|
| 51 |
}
|