aaniket's picture
Upload tokenizer
d5a1162
raw
history blame
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 6,
"ae": 5,
"ah": 20,
"ao": 4,
"aw": 2,
"ax": 1,
"ax-h": 36,
"axr": 43,
"ay": 26,
"b": 40,
"bcl": 23,
"ch": 11,
"d": 51,
"dcl": 59,
"dh": 44,
"dx": 54,
"eh": 41,
"el": 47,
"em": 53,
"en": 56,
"eng": 14,
"epi": 48,
"er": 27,
"ey": 35,
"f": 24,
"g": 9,
"gcl": 12,
"h#": 60,
"hh": 18,
"hv": 28,
"ih": 38,
"ix": 3,
"iy": 33,
"jh": 58,
"k": 25,
"kcl": 0,
"l": 45,
"m": 37,
"n": 30,
"ng": 22,
"nx": 55,
"ow": 7,
"oy": 49,
"p": 52,
"pau": 39,
"pcl": 57,
"q": 29,
"r": 10,
"s": 32,
"sh": 46,
"t": 50,
"tcl": 31,
"th": 13,
"uh": 19,
"uw": 21,
"ux": 34,
"v": 16,
"w": 8,
"y": 15,
"z": 17,
"zh": 42,
"|": 61
}