Upload tokenizer
Browse files- vocab.json +127 -127
vocab.json
CHANGED
@@ -1,133 +1,133 @@
|
|
1 |
{
|
2 |
-
"'":
|
3 |
-
"A":
|
4 |
-
"L":
|
5 |
-
"M":
|
6 |
"ML": 129,
|
7 |
-
"T":
|
8 |
"TA": 130,
|
9 |
-
"[":
|
10 |
"[PAD]": 128,
|
11 |
"[UNK]": 127,
|
12 |
-
"]":
|
13 |
-
"|":
|
14 |
-
"ஃ":
|
15 |
-
"அ":
|
16 |
-
"ஆ":
|
17 |
-
"இ":
|
18 |
-
"ஈ":
|
19 |
-
"உ":
|
20 |
-
"ஊ":
|
21 |
-
"எ":
|
22 |
-
"ஏ":
|
23 |
-
"ஐ":
|
24 |
-
"ஒ":
|
25 |
-
"ஓ":
|
26 |
-
"க":
|
27 |
-
"ங":
|
28 |
-
"ச":
|
29 |
-
"ஜ":
|
30 |
-
"ஞ":
|
31 |
-
"ட":
|
32 |
-
"ண":
|
33 |
-
"த":
|
34 |
-
"ந":
|
35 |
-
"ன":
|
36 |
-
"ப":
|
37 |
-
"ம":
|
38 |
-
"ய":
|
39 |
-
"ர":
|
40 |
-
"ற":
|
41 |
-
"ல":
|
42 |
-
"ள":
|
43 |
-
"ழ":
|
44 |
-
"வ":
|
45 |
-
"ஷ":
|
46 |
-
"ஸ":
|
47 |
-
"ஹ":
|
48 |
-
"ா":
|
49 |
-
"ி":
|
50 |
-
"ீ":
|
51 |
-
"ு":
|
52 |
-
"ூ":
|
53 |
-
"ெ":
|
54 |
-
"ே":
|
55 |
-
"ை":
|
56 |
-
"ொ":
|
57 |
-
"ோ":
|
58 |
-
"ௌ":
|
59 |
-
"்":
|
60 |
-
"ം":
|
61 |
-
"ഃ":
|
62 |
-
"അ":
|
63 |
-
"ആ":
|
64 |
-
"ഇ":
|
65 |
-
"ഈ":
|
66 |
-
"ഉ":
|
67 |
-
"ഊ":
|
68 |
-
"എ":
|
69 |
-
"ഏ":
|
70 |
-
"ഐ":
|
71 |
-
"ഒ":
|
72 |
-
"ഓ":
|
73 |
-
"ക":
|
74 |
-
"ഖ":
|
75 |
-
"ഗ":
|
76 |
-
"ഘ":
|
77 |
-
"ങ":
|
78 |
-
"ച":
|
79 |
-
"ഛ":
|
80 |
-
"ജ":
|
81 |
-
"ഞ":
|
82 |
-
"ട":
|
83 |
-
"ഠ":
|
84 |
-
"ഡ":
|
85 |
-
"ഢ":
|
86 |
-
"ണ":
|
87 |
-
"ത":
|
88 |
-
"ഥ":
|
89 |
-
"ദ":
|
90 |
-
"ധ":
|
91 |
-
"ന":
|
92 |
-
"പ":
|
93 |
-
"ഫ":
|
94 |
-
"ബ":
|
95 |
-
"ഭ":
|
96 |
-
"മ":
|
97 |
-
"യ":
|
98 |
-
"ര":
|
99 |
-
"റ":
|
100 |
-
"ല":
|
101 |
-
"ള":
|
102 |
-
"ഴ":
|
103 |
-
"വ":
|
104 |
-
"ശ":
|
105 |
-
"ഷ":
|
106 |
-
"സ":
|
107 |
-
"ഹ":
|
108 |
-
"ാ":
|
109 |
-
"ി":
|
110 |
-
"ീ":
|
111 |
-
"ു":
|
112 |
-
"ൂ":
|
113 |
-
"ൃ":
|
114 |
-
"െ":
|
115 |
-
"േ":
|
116 |
-
"ൈ":
|
117 |
-
"ൊ":
|
118 |
-
"ോ":
|
119 |
-
"ൌ":
|
120 |
-
"്":
|
121 |
-
"ൗ":
|
122 |
-
"ൺ":
|
123 |
-
"ൻ":
|
124 |
-
"ർ":
|
125 |
-
"ൽ":
|
126 |
-
"ൾ":
|
127 |
-
"ൿ":
|
128 |
-
"‘":
|
129 |
-
"’":
|
130 |
-
"“":
|
131 |
-
"”":
|
132 |
-
"●":
|
133 |
}
|
|
|
1 |
{
|
2 |
+
"'": 34,
|
3 |
+
"A": 99,
|
4 |
+
"L": 92,
|
5 |
+
"M": 78,
|
6 |
"ML": 129,
|
7 |
+
"T": 18,
|
8 |
"TA": 130,
|
9 |
+
"[": 106,
|
10 |
"[PAD]": 128,
|
11 |
"[UNK]": 127,
|
12 |
+
"]": 72,
|
13 |
+
"|": 47,
|
14 |
+
"ஃ": 57,
|
15 |
+
"அ": 45,
|
16 |
+
"ஆ": 83,
|
17 |
+
"இ": 108,
|
18 |
+
"ஈ": 10,
|
19 |
+
"உ": 29,
|
20 |
+
"ஊ": 113,
|
21 |
+
"எ": 11,
|
22 |
+
"ஏ": 103,
|
23 |
+
"ஐ": 8,
|
24 |
+
"ஒ": 84,
|
25 |
+
"ஓ": 116,
|
26 |
+
"க": 73,
|
27 |
+
"ங": 89,
|
28 |
+
"ச": 53,
|
29 |
+
"ஜ": 62,
|
30 |
+
"ஞ": 41,
|
31 |
+
"ட": 51,
|
32 |
+
"ண": 44,
|
33 |
+
"த": 111,
|
34 |
+
"ந": 28,
|
35 |
+
"ன": 98,
|
36 |
+
"ப": 69,
|
37 |
+
"ம": 5,
|
38 |
+
"ய": 87,
|
39 |
+
"ர": 49,
|
40 |
+
"ற": 105,
|
41 |
+
"ல": 59,
|
42 |
+
"ள": 50,
|
43 |
+
"ழ": 82,
|
44 |
+
"வ": 115,
|
45 |
+
"ஷ": 94,
|
46 |
+
"ஸ": 104,
|
47 |
+
"ஹ": 81,
|
48 |
+
"ா": 21,
|
49 |
+
"ி": 97,
|
50 |
+
"ீ": 70,
|
51 |
+
"ு": 65,
|
52 |
+
"ூ": 126,
|
53 |
+
"ெ": 90,
|
54 |
+
"ே": 122,
|
55 |
+
"ை": 38,
|
56 |
+
"ொ": 114,
|
57 |
+
"ோ": 27,
|
58 |
+
"ௌ": 102,
|
59 |
+
"்": 13,
|
60 |
+
"ം": 121,
|
61 |
+
"ഃ": 12,
|
62 |
+
"അ": 60,
|
63 |
+
"ആ": 23,
|
64 |
+
"ഇ": 31,
|
65 |
+
"ഈ": 68,
|
66 |
+
"ഉ": 30,
|
67 |
+
"ഊ": 26,
|
68 |
+
"എ": 95,
|
69 |
+
"ഏ": 67,
|
70 |
+
"ഐ": 0,
|
71 |
+
"ഒ": 64,
|
72 |
+
"ഓ": 112,
|
73 |
+
"ക": 76,
|
74 |
+
"ഖ": 100,
|
75 |
+
"ഗ": 109,
|
76 |
+
"ഘ": 75,
|
77 |
+
"ങ": 123,
|
78 |
+
"ച": 54,
|
79 |
+
"ഛ": 63,
|
80 |
+
"ജ": 9,
|
81 |
+
"ഞ": 119,
|
82 |
+
"ട": 3,
|
83 |
+
"ഠ": 74,
|
84 |
+
"ഡ": 16,
|
85 |
+
"ഢ": 37,
|
86 |
+
"ണ": 91,
|
87 |
+
"ത": 120,
|
88 |
+
"ഥ": 4,
|
89 |
+
"ദ": 7,
|
90 |
+
"ധ": 24,
|
91 |
+
"ന": 43,
|
92 |
+
"പ": 93,
|
93 |
+
"ഫ": 61,
|
94 |
+
"ബ": 125,
|
95 |
+
"ഭ": 14,
|
96 |
+
"മ": 17,
|
97 |
+
"യ": 58,
|
98 |
+
"ര": 101,
|
99 |
+
"റ": 86,
|
100 |
+
"ല": 2,
|
101 |
+
"ള": 117,
|
102 |
+
"ഴ": 35,
|
103 |
+
"വ": 85,
|
104 |
+
"ശ": 6,
|
105 |
+
"ഷ": 107,
|
106 |
+
"സ": 40,
|
107 |
+
"ഹ": 66,
|
108 |
+
"ാ": 42,
|
109 |
+
"ി": 36,
|
110 |
+
"ീ": 110,
|
111 |
+
"ു": 15,
|
112 |
+
"ൂ": 25,
|
113 |
+
"ൃ": 19,
|
114 |
+
"െ": 1,
|
115 |
+
"േ": 124,
|
116 |
+
"ൈ": 55,
|
117 |
+
"ൊ": 46,
|
118 |
+
"ോ": 77,
|
119 |
+
"ൌ": 56,
|
120 |
+
"്": 71,
|
121 |
+
"ൗ": 88,
|
122 |
+
"ൺ": 32,
|
123 |
+
"ൻ": 96,
|
124 |
+
"ർ": 39,
|
125 |
+
"ൽ": 33,
|
126 |
+
"ൾ": 48,
|
127 |
+
"ൿ": 52,
|
128 |
+
"‘": 79,
|
129 |
+
"’": 80,
|
130 |
+
"“": 20,
|
131 |
+
"”": 22,
|
132 |
+
"●": 118
|
133 |
}
|