Sania67's picture
add tokenizer
fb086ac
raw
history blame
268 Bytes
{"n": 0, "k": 1, "r": 2, "b": 3, "i": 4, "h": 5, "a": 7, "z": 8, "u": 9, "d": 10, "f": 11, "p": 12, "l": 13, "g": 14, "m": 15, "o": 16, "v": 17, "w": 18, "c": 19, "'": 20, "j": 21, "x": 22, "y": 23, "t": 24, "e": 25, "s": 26, "q": 27, "|": 6, "[UNK]": 28, "[PAD]": 29}