Upload tokenizer
Browse files- vocab.json +48 -48
vocab.json
CHANGED
@@ -1,53 +1,53 @@
|
|
1 |
{
|
2 |
-
"(":
|
3 |
-
")":
|
4 |
"[PAD]": 50,
|
5 |
"[UNK]": 49,
|
6 |
-
"|":
|
7 |
-
"«":
|
8 |
-
"»":
|
9 |
-
"՚":
|
10 |
-
"՛":
|
11 |
-
"՝":
|
12 |
-
"՞":
|
13 |
-
"ա":
|
14 |
-
"բ":
|
15 |
-
"գ":
|
16 |
-
"դ":
|
17 |
-
"ե":
|
18 |
-
"զ":
|
19 |
"է": 22,
|
20 |
-
"ը":
|
21 |
-
"թ":
|
22 |
-
"ժ":
|
23 |
-
"ի":
|
24 |
-
"լ":
|
25 |
-
"խ":
|
26 |
-
"ծ":
|
27 |
-
"կ":
|
28 |
-
"հ":
|
29 |
-
"ձ":
|
30 |
-
"ղ":
|
31 |
-
"ճ":
|
32 |
-
"մ":
|
33 |
-
"յ":
|
34 |
-
"ն":
|
35 |
-
"շ":
|
36 |
-
"ո":
|
37 |
-
"չ":
|
38 |
-
"պ":
|
39 |
-
"ջ":
|
40 |
-
"ռ":
|
41 |
-
"ս":
|
42 |
-
"վ":
|
43 |
-
"տ":
|
44 |
-
"ր":
|
45 |
-
"ց":
|
46 |
-
"ւ":
|
47 |
-
"փ":
|
48 |
-
"ք":
|
49 |
-
"օ":
|
50 |
-
"ֆ":
|
51 |
-
"և":
|
52 |
-
"։":
|
53 |
}
|
|
|
1 |
{
|
2 |
+
"(": 28,
|
3 |
+
")": 0,
|
4 |
"[PAD]": 50,
|
5 |
"[UNK]": 49,
|
6 |
+
"|": 1,
|
7 |
+
"«": 43,
|
8 |
+
"»": 16,
|
9 |
+
"՚": 37,
|
10 |
+
"՛": 30,
|
11 |
+
"՝": 42,
|
12 |
+
"՞": 11,
|
13 |
+
"ա": 46,
|
14 |
+
"բ": 13,
|
15 |
+
"գ": 24,
|
16 |
+
"դ": 5,
|
17 |
+
"ե": 32,
|
18 |
+
"զ": 26,
|
19 |
"է": 22,
|
20 |
+
"ը": 44,
|
21 |
+
"թ": 47,
|
22 |
+
"ժ": 39,
|
23 |
+
"ի": 45,
|
24 |
+
"լ": 48,
|
25 |
+
"խ": 4,
|
26 |
+
"ծ": 33,
|
27 |
+
"կ": 20,
|
28 |
+
"հ": 41,
|
29 |
+
"ձ": 31,
|
30 |
+
"ղ": 27,
|
31 |
+
"ճ": 29,
|
32 |
+
"մ": 15,
|
33 |
+
"յ": 14,
|
34 |
+
"ն": 35,
|
35 |
+
"շ": 21,
|
36 |
+
"ո": 40,
|
37 |
+
"չ": 25,
|
38 |
+
"պ": 17,
|
39 |
+
"ջ": 6,
|
40 |
+
"ռ": 38,
|
41 |
+
"ս": 12,
|
42 |
+
"վ": 9,
|
43 |
+
"տ": 23,
|
44 |
+
"ր": 18,
|
45 |
+
"ց": 3,
|
46 |
+
"ւ": 19,
|
47 |
+
"փ": 10,
|
48 |
+
"ք": 34,
|
49 |
+
"օ": 2,
|
50 |
+
"ֆ": 8,
|
51 |
+
"և": 7,
|
52 |
+
"։": 36
|
53 |
}
|