ad019el commited on
Commit
fb8c22c
1 Parent(s): 3dba7cd

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +45 -45
vocab.json CHANGED
@@ -1,50 +1,50 @@
1
  {
2
  "[PAD]": 47,
3
  "[UNK]": 46,
4
- "a": 31,
5
- "b": 11,
6
- "c": 36,
7
  "d": 21,
8
- "e": 28,
9
- "f": 41,
10
- "g": 30,
11
- "h": 19,
12
- "i": 5,
13
- "j": 43,
14
- "k": 40,
15
- "l": 14,
16
- "m": 37,
17
- "n": 29,
18
- "o": 25,
19
- "p": 26,
20
- "q": 8,
21
- "r": 4,
22
- "s": 0,
23
- "t": 38,
24
- "u": 12,
25
- "v": 2,
26
- "w": 20,
27
- "x": 13,
28
- "y": 9,
29
- "z": 6,
30
- "|": 33,
31
- " ": 42,
32
- "ĉ": 22,
33
- "č": 17,
34
- "ğ": 44,
35
- "ţ": 32,
36
- "ǧ": 7,
37
- "ɛ": 18,
38
- "ɣ": 1,
39
- "̣": 27,
40
- "γ": 10,
41
- "ε": 3,
42
- "ԑ": 34,
43
- "چ": 35,
44
- "ḍ": 16,
45
- "ḥ": 45,
46
- "ṛ": 15,
47
- "ṣ": 23,
48
- "ṭ": 39,
49
- "ẓ": 24
50
  }
 
1
  {
2
  "[PAD]": 47,
3
  "[UNK]": 46,
4
+ "a": 1,
5
+ "b": 22,
6
+ "c": 37,
7
  "d": 21,
8
+ "e": 2,
9
+ "f": 26,
10
+ "g": 34,
11
+ "h": 5,
12
+ "i": 44,
13
+ "j": 45,
14
+ "k": 28,
15
+ "l": 9,
16
+ "m": 18,
17
+ "n": 36,
18
+ "o": 38,
19
+ "p": 10,
20
+ "q": 40,
21
+ "r": 11,
22
+ "s": 33,
23
+ "t": 6,
24
+ "u": 24,
25
+ "v": 41,
26
+ "w": 32,
27
+ "x": 20,
28
+ "y": 7,
29
+ "z": 29,
30
+ "|": 15,
31
+ " ": 30,
32
+ "ĉ": 42,
33
+ "č": 27,
34
+ "ğ": 3,
35
+ "ţ": 0,
36
+ "ǧ": 14,
37
+ "ɛ": 25,
38
+ "ɣ": 43,
39
+ "̣": 35,
40
+ "γ": 4,
41
+ "ε": 39,
42
+ "ԑ": 23,
43
+ "چ": 19,
44
+ "ḍ": 8,
45
+ "ḥ": 12,
46
+ "ṛ": 31,
47
+ "ṣ": 16,
48
+ "ṭ": 17,
49
+ "ẓ": 13
50
  }