{ "[PAD]": 0, "[UNK]": 1, "aa": 3, "aai": 4, "aak": 5, "aam": 6, "aan": 7, "aang": 8, "aap": 9, "aat": 10, "aau": 11, "ai": 12, "ak": 13, "am": 14, "an": 15, "ang": 16, "ap": 17, "at": 18, "au": 19, "b": 20, "c": 21, "d": 22, "e": 23, "ei": 24, "ek": 25, "eng": 26, "eoi": 27, "eon": 28, "eot": 29, "ep": 30, "eu": 31, "f": 32, "g": 33, "gw": 34, "h": 35, "i": 36, "ik": 37, "im": 38, "in": 39, "ing": 40, "ip": 41, "it": 42, "iu": 43, "j": 44, "k": 45, "kw": 46, "l": 47, "m": 48, "n": 49, "ng": 50, "o": 51, "oe": 52, "oek": 53, "oeng": 54, "oi": 55, "ok": 56, "on": 57, "ong": 58, "ot": 59, "ou": 60, "p": 61, "s": 62, "t": 63, "u": 64, "ui": 65, "uk": 66, "un": 67, "ung": 68, "ut": 69, "w": 70, "yu": 71, "yun": 72, "yut": 73, "z": 74, "|": 2 }