aaniket's picture
Upload tokenizer
6b2be5a
raw
history blame
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 37,
"ae": 38,
"ah": 16,
"ao": 9,
"aw": 34,
"ax": 49,
"ax-h": 15,
"axr": 5,
"ay": 27,
"b": 4,
"bcl": 1,
"ch": 45,
"d": 31,
"dcl": 18,
"dh": 8,
"dx": 19,
"eh": 12,
"el": 44,
"em": 22,
"en": 7,
"eng": 33,
"epi": 57,
"er": 13,
"ey": 20,
"f": 29,
"g": 59,
"gcl": 0,
"h#": 21,
"hh": 25,
"hv": 60,
"ih": 42,
"ix": 41,
"iy": 53,
"jh": 51,
"k": 40,
"kcl": 43,
"l": 3,
"m": 26,
"n": 52,
"ng": 39,
"nx": 6,
"ow": 35,
"oy": 2,
"p": 56,
"pau": 14,
"pcl": 48,
"q": 58,
"r": 30,
"s": 55,
"sh": 10,
"t": 54,
"tcl": 47,
"th": 36,
"uh": 24,
"uw": 46,
"ux": 11,
"v": 17,
"w": 23,
"y": 32,
"z": 50,
"zh": 28,
"|": 61
}