jessiejohnson's picture
add tokenizer
745e058
raw
history blame
620 Bytes
{"uh": 0, "er": 1, "nx": 2, "sh": 3, "dh": 4, "oy": 5, "w": 6, "tcl": 7, "ow": 8, "en": 9, "ax-h": 10, "ix": 11, "epi": 12, "b": 13, "r": 14, "ch": 15, "ih": 16, "q": 17, "hv": 18, "ax": 19, "zh": 20, "m": 21, "em": 22, "jh": 23, "s": 24, "hh": 25, "axr": 26, "ey": 27, "th": 28, "gcl": 29, "d": 30, "g": 31, "ux": 32, "p": 33, "y": 34, "pau": 35, "v": 36, "kcl": 37, "uw": 38, "ay": 39, "ng": 40, "f": 41, "ae": 42, "z": 43, "dx": 44, "pcl": 45, "iy": 46, "l": 47, "eng": 48, "ah": 49, "t": 50, "ao": 51, "el": 52, "bcl": 53, "aa": 54, "aw": 55, "k": 56, "n": 57, "dcl": 58, "eh": 59, "|": 60, "[UNK]": 61, "[PAD]": 62}