ad019el commited on
Commit
611b3e2
1 Parent(s): 1ea1168

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