Commit
·
cfa4c91
1
Parent(s):
d8f6149
Update vocab.json
Browse files- vocab.json +71 -81
vocab.json
CHANGED
|
@@ -1,88 +1,78 @@
|
|
| 1 |
{
|
| 2 |
-
"0": 5,
|
| 3 |
-
"1": 6,
|
| 4 |
-
"2": 7,
|
| 5 |
-
"3": 8,
|
| 6 |
-
"4": 9,
|
| 7 |
-
"5": 10,
|
| 8 |
-
"6": 11,
|
| 9 |
-
"7": 12,
|
| 10 |
-
"8": 13,
|
| 11 |
-
"9": 14,
|
| 12 |
"</s>": 2,
|
| 13 |
"<pad>": 0,
|
| 14 |
"<s>": 1,
|
| 15 |
"<unk>": 3,
|
| 16 |
"|": 4,
|
| 17 |
-
"ँ":
|
| 18 |
-
"ं":
|
| 19 |
-
"ः":
|
| 20 |
-
"अ":
|
| 21 |
-
"आ":
|
| 22 |
-
"इ":
|
| 23 |
-
"ई":
|
| 24 |
-
"उ":
|
| 25 |
-
"ऊ":
|
| 26 |
-
"ऋ":
|
| 27 |
-
"ए":
|
| 28 |
-
"ऐ":
|
| 29 |
-
"ऑ":
|
| 30 |
-
"ओ":
|
| 31 |
-
"औ":
|
| 32 |
-
"क":
|
| 33 |
-
"ख":
|
| 34 |
-
"ग":
|
| 35 |
-
"घ":
|
| 36 |
-
"ङ":
|
| 37 |
-
"च":
|
| 38 |
-
"छ":
|
| 39 |
-
"ज":
|
| 40 |
-
"झ":
|
| 41 |
-
"ञ":
|
| 42 |
-
"ट":
|
| 43 |
-
"ठ":
|
| 44 |
-
"ड":
|
| 45 |
-
"ढ":
|
| 46 |
-
"ण":
|
| 47 |
-
"त":
|
| 48 |
-
"थ":
|
| 49 |
-
"द":
|
| 50 |
-
"ध":
|
| 51 |
-
"न":
|
| 52 |
-
"प":
|
| 53 |
-
"फ":
|
| 54 |
-
"ब":
|
| 55 |
-
"भ":
|
| 56 |
-
"म":
|
| 57 |
-
"य":
|
| 58 |
-
"र":
|
| 59 |
-
"ल":
|
| 60 |
-
"व":
|
| 61 |
-
"श":
|
| 62 |
-
"ष":
|
| 63 |
-
"स":
|
| 64 |
-
"ह":
|
| 65 |
-
"ा":
|
| 66 |
-
"ि":
|
| 67 |
-
"ी":
|
| 68 |
-
"ु":
|
| 69 |
-
"ू":
|
| 70 |
-
"ृ":
|
| 71 |
-
"ॅ":
|
| 72 |
-
"े":
|
| 73 |
-
"ै":
|
| 74 |
-
"ॉ":
|
| 75 |
-
"ो":
|
| 76 |
-
"ौ":
|
| 77 |
-
"्":
|
| 78 |
-
"क़":
|
| 79 |
-
"ख़":
|
| 80 |
-
"ग़":
|
| 81 |
-
"ज़":
|
| 82 |
-
"ड़":
|
| 83 |
-
"ढ़":
|
| 84 |
-
"फ़":
|
| 85 |
-
"य़":
|
| 86 |
-
"।":
|
| 87 |
-
"़":
|
| 88 |
}
|
|
|
|
| 1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
"</s>": 2,
|
| 3 |
"<pad>": 0,
|
| 4 |
"<s>": 1,
|
| 5 |
"<unk>": 3,
|
| 6 |
"|": 4,
|
| 7 |
+
"ँ": 5,
|
| 8 |
+
"ं": 6,
|
| 9 |
+
"ः": 7,
|
| 10 |
+
"अ": 8,
|
| 11 |
+
"आ": 9,
|
| 12 |
+
"इ": 10,
|
| 13 |
+
"ई": 11,
|
| 14 |
+
"उ": 12,
|
| 15 |
+
"ऊ": 13,
|
| 16 |
+
"ऋ": 14,
|
| 17 |
+
"ए": 15,
|
| 18 |
+
"ऐ": 16,
|
| 19 |
+
"ऑ": 17,
|
| 20 |
+
"ओ": 18,
|
| 21 |
+
"औ": 19,
|
| 22 |
+
"क": 20,
|
| 23 |
+
"ख": 21,
|
| 24 |
+
"ग": 22,
|
| 25 |
+
"घ": 23,
|
| 26 |
+
"ङ": 24,
|
| 27 |
+
"च": 25,
|
| 28 |
+
"छ": 26,
|
| 29 |
+
"ज": 27,
|
| 30 |
+
"झ": 28,
|
| 31 |
+
"ञ": 29,
|
| 32 |
+
"ट": 30,
|
| 33 |
+
"ठ": 31,
|
| 34 |
+
"ड": 32,
|
| 35 |
+
"ढ": 33,
|
| 36 |
+
"ण": 34,
|
| 37 |
+
"त": 35,
|
| 38 |
+
"थ": 36,
|
| 39 |
+
"द": 37,
|
| 40 |
+
"ध": 38,
|
| 41 |
+
"न": 39,
|
| 42 |
+
"प": 40,
|
| 43 |
+
"फ": 41,
|
| 44 |
+
"ब": 42,
|
| 45 |
+
"भ": 43,
|
| 46 |
+
"म": 44,
|
| 47 |
+
"य": 45,
|
| 48 |
+
"र": 46,
|
| 49 |
+
"ल": 47,
|
| 50 |
+
"व": 48,
|
| 51 |
+
"श": 49,
|
| 52 |
+
"ष": 50,
|
| 53 |
+
"स": 51,
|
| 54 |
+
"ह": 52,
|
| 55 |
+
"ा": 53,
|
| 56 |
+
"ि": 54,
|
| 57 |
+
"ी": 55,
|
| 58 |
+
"ु": 56,
|
| 59 |
+
"ू": 57,
|
| 60 |
+
"ृ": 58,
|
| 61 |
+
"ॅ": 59,
|
| 62 |
+
"े": 50,
|
| 63 |
+
"ै": 61,
|
| 64 |
+
"ॉ": 62,
|
| 65 |
+
"ो": 63,
|
| 66 |
+
"ौ": 64,
|
| 67 |
+
"्": 65,
|
| 68 |
+
"क़": 66,
|
| 69 |
+
"ख़": 67,
|
| 70 |
+
"ग़": 68,
|
| 71 |
+
"ज़": 69,
|
| 72 |
+
"ड़": 70,
|
| 73 |
+
"ढ़": 71,
|
| 74 |
+
"फ़": 72,
|
| 75 |
+
"य़": 73,
|
| 76 |
+
"।": 74,
|
| 77 |
+
"़": 75
|
| 78 |
}
|