aaniket's picture
Upload tokenizer
c310ef1
raw
history blame
No virus
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 36,
"ae": 57,
"ah": 10,
"ao": 25,
"aw": 26,
"ax": 56,
"ax-h": 32,
"axr": 49,
"ay": 60,
"b": 19,
"bcl": 59,
"ch": 0,
"d": 39,
"dcl": 9,
"dh": 48,
"dx": 58,
"eh": 45,
"el": 6,
"em": 16,
"en": 35,
"eng": 1,
"epi": 5,
"er": 38,
"ey": 52,
"f": 27,
"g": 47,
"gcl": 29,
"h#": 33,
"hh": 30,
"hv": 44,
"ih": 11,
"ix": 50,
"iy": 22,
"jh": 12,
"k": 8,
"kcl": 23,
"l": 53,
"m": 42,
"n": 21,
"ng": 41,
"nx": 17,
"ow": 46,
"oy": 37,
"p": 7,
"pau": 54,
"pcl": 55,
"q": 34,
"r": 31,
"s": 18,
"sh": 24,
"t": 40,
"tcl": 3,
"th": 13,
"uh": 20,
"uw": 4,
"ux": 43,
"v": 28,
"w": 51,
"y": 14,
"z": 15,
"zh": 2,
"|": 61
}