Upload tokenizer
364c947
|
{ |
|
"&": 12, |
|
"'": 20, |
|
"[PAD]": 101, |
|
"[UNK]": 100, |
|
"a": 40, |
|
"b": 33, |
|
"c": 65, |
|
"d": 41, |
|
"e": 84, |
|
"f": 14, |
|
"g": 26, |
|
"h": 13, |
|
"i": 87, |
|
"j": 91, |
|
"k": 28, |
|
"l": 86, |
|
"m": 53, |
|
"n": 5, |
|
"o": 7, |
|
"p": 39, |
|
"q": 30, |
|
"r": 19, |
|
"s": 71, |
|
"t": 75, |
|
"u": 17, |
|
"v": 46, |
|
"w": 90, |
|
"x": 16, |
|
"y": 10, |
|
"z": 97, |
|
"|": 43, |
|
"ँ": 32, |
|
"ं": 23, |
|
"ः": 38, |
|
"अ": 42, |
|
"आ": 2, |
|
"इ": 36, |
|
"ई": 80, |
|
"उ": 62, |
|
"ऊ": 100, |
|
"ऋ": 24, |
|
"ए": 98, |
|
"ऐ": 55, |
|
"ऑ": 47, |
|
"ओ": 72, |
|
"औ": 18, |
|
"क": 63, |
|
"ख": 1, |
|
"ग": 9, |
|
"घ": 15, |
|
"च": 45, |
|
"छ": 6, |
|
"ज": 64, |
|
"झ": 11, |
|
"ञ": 4, |
|
"ट": 34, |
|
"ठ": 25, |
|
"ड": 59, |
|
"ढ": 37, |
|
"ण": 27, |
|
"त": 31, |
|
"थ": 88, |
|
"द": 48, |
|
"ध": 8, |
|
"न": 95, |
|
"प": 92, |
|
"फ": 57, |
|
"ब": 68, |
|
"भ": 99, |
|
"म": 54, |
|
"य": 73, |
|
"र": 61, |
|
"ल": 79, |
|
"व": 85, |
|
"श": 22, |
|
"ष": 96, |
|
"स": 51, |
|
"ह": 50, |
|
"़": 44, |
|
"ा": 66, |
|
"ि": 74, |
|
"ी": 49, |
|
"ु": 70, |
|
"ू": 89, |
|
"ृ": 69, |
|
"ॅ": 76, |
|
"े": 21, |
|
"ै": 0, |
|
"ॉ": 58, |
|
"ो": 67, |
|
"ौ": 94, |
|
"्": 35, |
|
"क़": 56, |
|
"ख़": 78, |
|
"ग़": 82, |
|
"ज़": 29, |
|
"ड़": 83, |
|
"ढ़": 77, |
|
"फ़": 93, |
|
"।": 3, |
|
"–": 52, |
|
"’": 81 |
|
} |
|
|