Upload tokenizer
Browse files- vocab.json +57 -57
vocab.json
CHANGED
|
@@ -1,64 +1,64 @@
|
|
| 1 |
{
|
| 2 |
"[PAD]": 61,
|
| 3 |
"[UNK]": 60,
|
| 4 |
-
"|":
|
| 5 |
-
"ँ":
|
| 6 |
-
"ं":
|
| 7 |
-
"ः":
|
| 8 |
-
"अ":
|
| 9 |
-
"आ":
|
| 10 |
"इ": 26,
|
| 11 |
-
"ई":
|
| 12 |
-
"उ":
|
| 13 |
-
"ऊ":
|
| 14 |
-
"ए":
|
| 15 |
-
"ओ":
|
| 16 |
-
"औ":
|
| 17 |
-
"क":
|
| 18 |
-
"ख":
|
| 19 |
"ग": 12,
|
| 20 |
-
"घ":
|
| 21 |
-
"ङ":
|
| 22 |
-
"च":
|
| 23 |
-
"छ":
|
| 24 |
-
"ज":
|
| 25 |
-
"झ":
|
| 26 |
-
"ञ":
|
| 27 |
-
"ट":
|
| 28 |
-
"ठ":
|
| 29 |
-
"ड":
|
| 30 |
-
"ढ":
|
| 31 |
-
"ण":
|
| 32 |
-
"त":
|
| 33 |
-
"थ":
|
| 34 |
-
"द":
|
| 35 |
-
"ध":
|
| 36 |
-
"न":
|
| 37 |
-
"प":
|
| 38 |
-
"फ":
|
| 39 |
-
"ब":
|
| 40 |
-
"भ":
|
| 41 |
-
"म":
|
| 42 |
"य": 49,
|
| 43 |
-
"र":
|
| 44 |
-
"ल":
|
| 45 |
-
"व":
|
| 46 |
-
"श":
|
| 47 |
-
"ष":
|
| 48 |
-
"स":
|
| 49 |
-
"ह":
|
| 50 |
-
"ा":
|
| 51 |
-
"ि":
|
| 52 |
-
"ी":
|
| 53 |
-
"ु":
|
| 54 |
-
"ू":
|
| 55 |
-
"ृ":
|
| 56 |
-
"े":
|
| 57 |
-
"ै":
|
| 58 |
-
"ो":
|
| 59 |
-
"ौ":
|
| 60 |
-
"्":
|
| 61 |
-
"।":
|
| 62 |
-
"–":
|
| 63 |
-
"’":
|
| 64 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"[PAD]": 61,
|
| 3 |
"[UNK]": 60,
|
| 4 |
+
"|": 47,
|
| 5 |
+
"ँ": 24,
|
| 6 |
+
"ं": 48,
|
| 7 |
+
"ः": 25,
|
| 8 |
+
"अ": 42,
|
| 9 |
+
"आ": 4,
|
| 10 |
"इ": 26,
|
| 11 |
+
"ई": 14,
|
| 12 |
+
"उ": 10,
|
| 13 |
+
"ऊ": 13,
|
| 14 |
+
"ए": 27,
|
| 15 |
+
"ओ": 30,
|
| 16 |
+
"औ": 11,
|
| 17 |
+
"क": 40,
|
| 18 |
+
"ख": 39,
|
| 19 |
"ग": 12,
|
| 20 |
+
"घ": 36,
|
| 21 |
+
"ङ": 41,
|
| 22 |
+
"च": 57,
|
| 23 |
+
"छ": 6,
|
| 24 |
+
"ज": 9,
|
| 25 |
+
"झ": 44,
|
| 26 |
+
"ञ": 1,
|
| 27 |
+
"ट": 20,
|
| 28 |
+
"ठ": 2,
|
| 29 |
+
"ड": 28,
|
| 30 |
+
"ढ": 50,
|
| 31 |
+
"ण": 5,
|
| 32 |
+
"त": 43,
|
| 33 |
+
"थ": 22,
|
| 34 |
+
"द": 32,
|
| 35 |
+
"ध": 31,
|
| 36 |
+
"न": 55,
|
| 37 |
+
"प": 46,
|
| 38 |
+
"फ": 45,
|
| 39 |
+
"ब": 53,
|
| 40 |
+
"भ": 15,
|
| 41 |
+
"म": 33,
|
| 42 |
"य": 49,
|
| 43 |
+
"र": 21,
|
| 44 |
+
"ल": 59,
|
| 45 |
+
"व": 7,
|
| 46 |
+
"श": 8,
|
| 47 |
+
"ष": 38,
|
| 48 |
+
"स": 54,
|
| 49 |
+
"ह": 19,
|
| 50 |
+
"ा": 52,
|
| 51 |
+
"ि": 3,
|
| 52 |
+
"ी": 16,
|
| 53 |
+
"ु": 58,
|
| 54 |
+
"ू": 29,
|
| 55 |
+
"ृ": 56,
|
| 56 |
+
"े": 17,
|
| 57 |
+
"ै": 37,
|
| 58 |
+
"ो": 23,
|
| 59 |
+
"ौ": 18,
|
| 60 |
+
"्": 34,
|
| 61 |
+
"।": 0,
|
| 62 |
+
"–": 51,
|
| 63 |
+
"’": 35
|
| 64 |
}
|