aaniket's picture
Upload tokenizer
6ceadf0
raw
history blame
761 Bytes
{
" ": 61,
"[PAD]": 63,
"[UNK]": 62,
"aa": 35,
"ae": 29,
"ah": 55,
"ao": 33,
"aw": 41,
"ax": 21,
"ax-h": 14,
"axr": 26,
"ay": 11,
"b": 5,
"bcl": 10,
"ch": 1,
"d": 19,
"dcl": 49,
"dh": 38,
"dx": 48,
"eh": 52,
"el": 0,
"em": 17,
"en": 51,
"eng": 12,
"epi": 54,
"er": 50,
"ey": 8,
"f": 25,
"g": 56,
"gcl": 32,
"h#": 3,
"hh": 44,
"hv": 7,
"ih": 47,
"ix": 37,
"iy": 58,
"jh": 40,
"k": 15,
"kcl": 34,
"l": 13,
"m": 16,
"n": 2,
"ng": 9,
"nx": 57,
"ow": 53,
"oy": 45,
"p": 42,
"pau": 46,
"pcl": 18,
"q": 20,
"r": 4,
"s": 31,
"sh": 36,
"t": 43,
"tcl": 6,
"th": 22,
"uh": 30,
"uw": 23,
"ux": 39,
"v": 24,
"w": 59,
"y": 27,
"z": 28,
"zh": 60
}