sulaimank commited on
Commit
2e65465
·
verified ·
1 Parent(s): f5132d4

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +41 -35
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 38,
3
- "<s>": 37
4
  }
 
1
  {
2
+ "</s>": 44,
3
+ "<s>": 43
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "35": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "36": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "37": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "38": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "41": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "42": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "43": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "44": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,41 +1,47 @@
1
  {
2
  "lug": {
3
  "'": 1,
4
- "[PAD]": 36,
5
- "[UNK]": 35,
6
- "a": 2,
7
- "b": 3,
8
- "c": 4,
9
- "d": 5,
10
- "e": 6,
11
- "f": 7,
12
- "g": 8,
13
- "h": 9,
14
- "i": 10,
15
- "j": 11,
16
- "k": 12,
17
- "l": 13,
18
- "m": 14,
19
- "n": 15,
20
- "o": 16,
21
- "p": 17,
22
- "q": 18,
23
- "r": 19,
24
- "s": 20,
25
- "t": 21,
26
- "u": 22,
27
- "v": 23,
28
- "w": 24,
29
- "x": 25,
30
- "y": 26,
31
- "z": 27,
 
 
 
 
 
 
32
  "|": 0,
33
- "á": 28,
34
- "é": 29,
35
- "ó": 30,
36
- "ŋ": 31,
37
- "š": 32,
38
- "η": 33,
39
- "ղ": 34
40
  }
41
  }
 
1
  {
2
  "lug": {
3
  "'": 1,
4
+ "0": 2,
5
+ "1": 3,
6
+ "3": 4,
7
+ "5": 5,
8
+ "7": 6,
9
+ "9": 7,
10
+ "[PAD]": 42,
11
+ "[UNK]": 41,
12
+ "a": 8,
13
+ "b": 9,
14
+ "c": 10,
15
+ "d": 11,
16
+ "e": 12,
17
+ "f": 13,
18
+ "g": 14,
19
+ "h": 15,
20
+ "i": 16,
21
+ "j": 17,
22
+ "k": 18,
23
+ "l": 19,
24
+ "m": 20,
25
+ "n": 21,
26
+ "o": 22,
27
+ "p": 23,
28
+ "q": 24,
29
+ "r": 25,
30
+ "s": 26,
31
+ "t": 27,
32
+ "u": 28,
33
+ "v": 29,
34
+ "w": 30,
35
+ "x": 31,
36
+ "y": 32,
37
+ "z": 33,
38
  "|": 0,
39
+ "á": 34,
40
+ "é": 35,
41
+ "ó": 36,
42
+ "ŋ": 37,
43
+ "š": 38,
44
+ "η": 39,
45
+ "ղ": 40
46
  }
47
  }