nikhil6041's picture
add tokenizer
8e680bb
raw
history blame contribute delete
784 Bytes
{"इ": 0, "द": 1, "ण": 2, "l": 3, "ि": 4, "थ": 5, "m": 6, "ू": 7, "भ": 8, "ढ": 9, "ब": 10, "क": 11, "ध": 12, "ओ": 13, "ल": 14, "अ": 15, "व": 16, "े": 17, "उ": 18, "ी": 19, "a": 20, "p": 21, "e": 22, "ृ": 23, "स": 24, "छ": 25, "श": 26, "ग": 27, "य": 28, "ऑ": 29, "f": 30, "आ": 32, "त": 33, "च": 34, "ष": 35, "ा": 36, "ॉ": 37, "u": 38, "ज": 39, "ै": 40, "ं": 41, "'": 42, "ऐ": 43, "ः": 44, "न": 45, "औ": 46, "ठ": 47, "ए": 48, "ई": 49, "w": 50, "i": 51, "प": 52, "झ": 53, "ु": 54, "ौ": 55, "र": 56, "ह": 57, "म": 58, "ड": 59, "ऊ": 60, "ट": 61, "घ": 62, "्": 63, "ड़": 64, "ँ": 65, "।": 66, "ो": 67, "फ": 68, "r": 69, "़": 70, "ख": 71, "|": 31, "[UNK]": 72, "[PAD]": 73}