aaniket's picture
Upload tokenizer
c2e3075
raw
history blame
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 24,
"ae": 6,
"ah": 12,
"ao": 40,
"aw": 35,
"ax": 15,
"ax-h": 28,
"axr": 47,
"ay": 55,
"b": 9,
"bcl": 54,
"ch": 36,
"d": 45,
"dcl": 52,
"dh": 21,
"dx": 37,
"eh": 0,
"el": 48,
"em": 59,
"en": 14,
"eng": 53,
"epi": 19,
"er": 25,
"ey": 16,
"f": 50,
"g": 46,
"gcl": 17,
"h#": 5,
"hh": 32,
"hv": 2,
"ih": 44,
"ix": 57,
"iy": 29,
"jh": 30,
"k": 3,
"kcl": 8,
"l": 1,
"m": 31,
"n": 42,
"ng": 43,
"nx": 60,
"ow": 27,
"oy": 49,
"p": 38,
"pau": 22,
"pcl": 10,
"q": 39,
"r": 33,
"s": 7,
"sh": 58,
"t": 41,
"tcl": 26,
"th": 13,
"uh": 4,
"uw": 56,
"ux": 23,
"v": 11,
"w": 18,
"y": 20,
"z": 34,
"zh": 51,
"|": 61
}