aaniket's picture
Upload tokenizer
7c79569
raw
history blame
750 Bytes
{
"[PAD]": 62,
"[UNK]": 61,
"aa": 47,
"ae": 29,
"ah": 44,
"ao": 37,
"aw": 30,
"ax": 41,
"ax-h": 33,
"axr": 13,
"ay": 12,
"b": 55,
"bcl": 56,
"ch": 46,
"d": 27,
"dcl": 2,
"dh": 17,
"dx": 0,
"eh": 1,
"el": 8,
"em": 39,
"en": 40,
"eng": 54,
"epi": 59,
"er": 5,
"ey": 60,
"f": 53,
"g": 45,
"gcl": 6,
"h#": 14,
"hh": 31,
"hv": 51,
"ih": 16,
"ix": 9,
"iy": 25,
"jh": 3,
"k": 48,
"kcl": 28,
"l": 23,
"m": 4,
"n": 38,
"ng": 21,
"nx": 49,
"ow": 35,
"oy": 24,
"p": 36,
"pau": 22,
"pcl": 34,
"q": 58,
"r": 42,
"s": 26,
"sh": 11,
"t": 18,
"tcl": 7,
"th": 43,
"uh": 57,
"uw": 10,
"ux": 15,
"v": 19,
"w": 20,
"y": 50,
"z": 52,
"zh": 32
}