Commit
·
ee39f68
1
Parent(s):
4f1f9e2
Upload tokenizer
Browse files- vocab.json +34 -34
vocab.json
CHANGED
@@ -1,44 +1,44 @@
|
|
1 |
{
|
2 |
"sah": {
|
3 |
-
"-":
|
4 |
"<pad>": 39,
|
5 |
"<unk>": 38,
|
6 |
-
"y":
|
7 |
-
"|":
|
8 |
-
"а":
|
9 |
-
"б":
|
10 |
-
"в":
|
11 |
-
"г":
|
12 |
-
"д":
|
13 |
"е": 29,
|
14 |
-
"ж":
|
15 |
-
"з":
|
16 |
-
"и":
|
17 |
-
"й":
|
18 |
-
"к":
|
19 |
-
"л":
|
20 |
-
"м":
|
21 |
-
"н":
|
22 |
-
"о":
|
23 |
-
"п":
|
24 |
"р": 23,
|
25 |
-
"с":
|
26 |
-
"т":
|
27 |
"у": 9,
|
28 |
-
"ф":
|
29 |
-
"х":
|
30 |
-
"ц":
|
31 |
-
"ч":
|
32 |
-
"ш":
|
33 |
-
"щ":
|
34 |
-
"ы":
|
35 |
-
"ь":
|
36 |
-
"э":
|
37 |
-
"я":
|
38 |
-
"ҕ":
|
39 |
-
"ҥ":
|
40 |
"ү": 5,
|
41 |
-
"һ":
|
42 |
-
"ө":
|
43 |
}
|
44 |
}
|
|
|
1 |
{
|
2 |
"sah": {
|
3 |
+
"-": 34,
|
4 |
"<pad>": 39,
|
5 |
"<unk>": 38,
|
6 |
+
"y": 18,
|
7 |
+
"|": 17,
|
8 |
+
"а": 2,
|
9 |
+
"б": 24,
|
10 |
+
"в": 0,
|
11 |
+
"г": 36,
|
12 |
+
"д": 15,
|
13 |
"е": 29,
|
14 |
+
"ж": 19,
|
15 |
+
"з": 20,
|
16 |
+
"и": 25,
|
17 |
+
"й": 16,
|
18 |
+
"к": 33,
|
19 |
+
"л": 35,
|
20 |
+
"м": 8,
|
21 |
+
"н": 27,
|
22 |
+
"о": 11,
|
23 |
+
"п": 13,
|
24 |
"р": 23,
|
25 |
+
"с": 12,
|
26 |
+
"т": 1,
|
27 |
"у": 9,
|
28 |
+
"ф": 32,
|
29 |
+
"х": 14,
|
30 |
+
"ц": 6,
|
31 |
+
"ч": 26,
|
32 |
+
"ш": 31,
|
33 |
+
"щ": 10,
|
34 |
+
"ы": 37,
|
35 |
+
"ь": 30,
|
36 |
+
"э": 4,
|
37 |
+
"я": 21,
|
38 |
+
"ҕ": 22,
|
39 |
+
"ҥ": 28,
|
40 |
"ү": 5,
|
41 |
+
"һ": 7,
|
42 |
+
"ө": 3
|
43 |
}
|
44 |
}
|