aaniket's picture
Upload tokenizer
1474b58
raw
history blame
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 5,
"ae": 59,
"ah": 4,
"ao": 15,
"aw": 33,
"ax": 10,
"ax-h": 58,
"axr": 12,
"ay": 41,
"b": 26,
"bcl": 3,
"ch": 24,
"d": 60,
"dcl": 35,
"dh": 8,
"dx": 16,
"eh": 19,
"el": 34,
"em": 42,
"en": 44,
"eng": 46,
"epi": 20,
"er": 22,
"ey": 48,
"f": 38,
"g": 14,
"gcl": 45,
"h#": 11,
"hh": 25,
"hv": 37,
"ih": 43,
"ix": 57,
"iy": 2,
"jh": 55,
"k": 30,
"kcl": 50,
"l": 21,
"m": 17,
"n": 47,
"ng": 31,
"nx": 13,
"ow": 49,
"oy": 53,
"p": 6,
"pau": 18,
"pcl": 51,
"q": 54,
"r": 39,
"s": 7,
"sh": 23,
"t": 36,
"tcl": 28,
"th": 27,
"uh": 1,
"uw": 29,
"ux": 32,
"v": 9,
"w": 0,
"y": 52,
"z": 40,
"zh": 56,
"|": 61
}