Upload tokenizer
Browse files- vocab.json +77 -77
vocab.json
CHANGED
|
@@ -1,84 +1,84 @@
|
|
| 1 |
{
|
| 2 |
"'": 77,
|
| 3 |
-
"L":
|
| 4 |
-
"M":
|
| 5 |
"ML": 80,
|
| 6 |
"TA": 81,
|
| 7 |
-
"[":
|
| 8 |
"[PAD]": 79,
|
| 9 |
"[UNK]": 78,
|
| 10 |
-
"]":
|
| 11 |
-
"|":
|
| 12 |
-
"ം":
|
| 13 |
-
"ഃ":
|
| 14 |
-
"അ":
|
| 15 |
-
"ആ":
|
| 16 |
-
"ഇ":
|
| 17 |
-
"ഈ":
|
| 18 |
-
"ഉ":
|
| 19 |
-
"ഊ":
|
| 20 |
-
"എ":
|
| 21 |
-
"ഏ":
|
| 22 |
-
"ഐ":
|
| 23 |
-
"ഒ":
|
| 24 |
-
"ഓ":
|
| 25 |
-
"ക":
|
| 26 |
-
"ഖ":
|
| 27 |
-
"ഗ":
|
| 28 |
-
"ഘ":
|
| 29 |
-
"ങ":
|
| 30 |
-
"ച":
|
| 31 |
-
"ഛ":
|
| 32 |
-
"ജ":
|
| 33 |
-
"ഞ":
|
| 34 |
-
"ട":
|
| 35 |
-
"ഠ":
|
| 36 |
-
"ഡ":
|
| 37 |
-
"ഢ":
|
| 38 |
-
"ണ":
|
| 39 |
-
"ത":
|
| 40 |
-
"ഥ":
|
| 41 |
-
"ദ":
|
| 42 |
-
"ധ":
|
| 43 |
-
"ന":
|
| 44 |
-
"പ":
|
| 45 |
-
"ഫ":
|
| 46 |
-
"ബ":
|
| 47 |
-
"ഭ":
|
| 48 |
-
"മ":
|
| 49 |
-
"യ":
|
| 50 |
-
"ര":
|
| 51 |
-
"റ":
|
| 52 |
-
"ല":
|
| 53 |
-
"ള":
|
| 54 |
-
"ഴ":
|
| 55 |
-
"വ":
|
| 56 |
-
"ശ":
|
| 57 |
-
"ഷ":
|
| 58 |
-
"സ":
|
| 59 |
-
"ഹ":
|
| 60 |
-
"ാ":
|
| 61 |
-
"ി":
|
| 62 |
-
"ീ":
|
| 63 |
-
"ു":
|
| 64 |
-
"ൂ":
|
| 65 |
-
"ൃ":
|
| 66 |
-
"െ":
|
| 67 |
-
"േ":
|
| 68 |
-
"ൈ":
|
| 69 |
-
"ൊ":
|
| 70 |
-
"ോ":
|
| 71 |
-
"ൌ":
|
| 72 |
-
"്":
|
| 73 |
-
"ൗ":
|
| 74 |
-
"ൺ":
|
| 75 |
-
"ൻ":
|
| 76 |
-
"ർ":
|
| 77 |
-
"ൽ":
|
| 78 |
-
"ൾ":
|
| 79 |
-
"ൿ":
|
| 80 |
-
"‘":
|
| 81 |
-
"’":
|
| 82 |
-
"“":
|
| 83 |
-
"”":
|
| 84 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"'": 77,
|
| 3 |
+
"L": 3,
|
| 4 |
+
"M": 15,
|
| 5 |
"ML": 80,
|
| 6 |
"TA": 81,
|
| 7 |
+
"[": 72,
|
| 8 |
"[PAD]": 79,
|
| 9 |
"[UNK]": 78,
|
| 10 |
+
"]": 55,
|
| 11 |
+
"|": 44,
|
| 12 |
+
"ം": 20,
|
| 13 |
+
"ഃ": 25,
|
| 14 |
+
"അ": 34,
|
| 15 |
+
"ആ": 63,
|
| 16 |
+
"ഇ": 30,
|
| 17 |
+
"ഈ": 35,
|
| 18 |
+
"ഉ": 21,
|
| 19 |
+
"ഊ": 27,
|
| 20 |
+
"എ": 51,
|
| 21 |
+
"ഏ": 37,
|
| 22 |
+
"ഐ": 58,
|
| 23 |
+
"ഒ": 46,
|
| 24 |
+
"ഓ": 62,
|
| 25 |
+
"ക": 65,
|
| 26 |
+
"ഖ": 69,
|
| 27 |
+
"ഗ": 0,
|
| 28 |
+
"ഘ": 17,
|
| 29 |
+
"ങ": 57,
|
| 30 |
+
"ച": 68,
|
| 31 |
+
"ഛ": 71,
|
| 32 |
+
"ജ": 76,
|
| 33 |
+
"ഞ": 47,
|
| 34 |
+
"ട": 14,
|
| 35 |
+
"ഠ": 74,
|
| 36 |
+
"ഡ": 39,
|
| 37 |
+
"ഢ": 1,
|
| 38 |
+
"ണ": 10,
|
| 39 |
+
"ത": 29,
|
| 40 |
+
"ഥ": 67,
|
| 41 |
+
"ദ": 18,
|
| 42 |
+
"ധ": 9,
|
| 43 |
+
"ന": 73,
|
| 44 |
+
"പ": 66,
|
| 45 |
+
"ഫ": 38,
|
| 46 |
+
"ബ": 61,
|
| 47 |
+
"ഭ": 45,
|
| 48 |
+
"മ": 19,
|
| 49 |
+
"യ": 16,
|
| 50 |
+
"ര": 7,
|
| 51 |
+
"റ": 59,
|
| 52 |
+
"ല": 28,
|
| 53 |
+
"ള": 40,
|
| 54 |
+
"ഴ": 2,
|
| 55 |
+
"വ": 54,
|
| 56 |
+
"ശ": 64,
|
| 57 |
+
"ഷ": 26,
|
| 58 |
+
"സ": 56,
|
| 59 |
+
"ഹ": 42,
|
| 60 |
+
"ാ": 5,
|
| 61 |
+
"ി": 49,
|
| 62 |
+
"ീ": 24,
|
| 63 |
+
"ു": 75,
|
| 64 |
+
"ൂ": 23,
|
| 65 |
+
"ൃ": 41,
|
| 66 |
+
"െ": 31,
|
| 67 |
+
"േ": 50,
|
| 68 |
+
"ൈ": 53,
|
| 69 |
+
"ൊ": 60,
|
| 70 |
+
"ോ": 48,
|
| 71 |
+
"ൌ": 70,
|
| 72 |
+
"്": 12,
|
| 73 |
+
"ൗ": 43,
|
| 74 |
+
"ൺ": 13,
|
| 75 |
+
"ൻ": 36,
|
| 76 |
+
"ർ": 6,
|
| 77 |
+
"ൽ": 32,
|
| 78 |
+
"ൾ": 52,
|
| 79 |
+
"ൿ": 11,
|
| 80 |
+
"‘": 8,
|
| 81 |
+
"’": 4,
|
| 82 |
+
"“": 33,
|
| 83 |
+
"”": 22
|
| 84 |
}
|