manjugeorge commited on
Commit
dbe45b0
1 Parent(s): e24d615

Upload tokenizer

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +6 -0
  2. tokenizer_config.json +10 -0
  3. vocab.json +78 -0
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "[PAD]",
5
+ "unk_token": "[UNK]"
6
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "do_lower_case": false,
4
+ "eos_token": "</s>",
5
+ "pad_token": "[PAD]",
6
+ "replace_word_delimiter_char": " ",
7
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
8
+ "unk_token": "[UNK]",
9
+ "word_delimiter_token": "|"
10
+ }
vocab.json ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "'": 23,
3
+ "[PAD]": 75,
4
+ "[UNK]": 74,
5
+ "|": 38,
6
+ "ം": 31,
7
+ "ഃ": 25,
8
+ "അ": 20,
9
+ "ആ": 8,
10
+ "ഇ": 61,
11
+ "ഈ": 48,
12
+ "ഉ": 73,
13
+ "ഊ": 27,
14
+ "എ": 68,
15
+ "ഏ": 63,
16
+ "ഐ": 64,
17
+ "ഒ": 40,
18
+ "ഓ": 22,
19
+ "ക": 32,
20
+ "ഖ": 59,
21
+ "ഗ": 37,
22
+ "ഘ": 60,
23
+ "ങ": 3,
24
+ "ച": 65,
25
+ "ഛ": 5,
26
+ "ജ": 11,
27
+ "ഞ": 17,
28
+ "ട": 58,
29
+ "ഠ": 30,
30
+ "ഡ": 1,
31
+ "ഢ": 33,
32
+ "ണ": 0,
33
+ "ത": 45,
34
+ "ഥ": 44,
35
+ "ദ": 42,
36
+ "ധ": 16,
37
+ "ന": 12,
38
+ "പ": 66,
39
+ "ഫ": 56,
40
+ "ബ": 50,
41
+ "ഭ": 9,
42
+ "മ": 36,
43
+ "യ": 13,
44
+ "ര": 51,
45
+ "റ": 26,
46
+ "ല": 41,
47
+ "ള": 47,
48
+ "ഴ": 62,
49
+ "വ": 29,
50
+ "ശ": 52,
51
+ "ഷ": 70,
52
+ "സ": 54,
53
+ "ഹ": 49,
54
+ "ാ": 34,
55
+ "ി": 21,
56
+ "ീ": 4,
57
+ "ു": 67,
58
+ "ൂ": 57,
59
+ "ൃ": 2,
60
+ "െ": 43,
61
+ "േ": 19,
62
+ "ൈ": 39,
63
+ "ൊ": 15,
64
+ "ോ": 14,
65
+ "ൌ": 53,
66
+ "്": 18,
67
+ "ൗ": 28,
68
+ "ൺ": 35,
69
+ "ൻ": 69,
70
+ "ർ": 46,
71
+ "ൽ": 6,
72
+ "ൾ": 7,
73
+ "ൿ": 24,
74
+ "‘": 10,
75
+ "’": 72,
76
+ "“": 55,
77
+ "”": 71
78
+ }