aaniket's picture
Upload tokenizer
9e00df1
raw history blame
No virus
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 36,
"ae": 47,
"ah": 55,
"ao": 9,
"aw": 54,
"ax": 27,
"ax-h": 34,
"axr": 50,
"ay": 32,
"b": 24,
"bcl": 56,
"ch": 6,
"d": 37,
"dcl": 45,
"dh": 3,
"dx": 5,
"eh": 15,
"el": 51,
"em": 48,
"en": 46,
"eng": 31,
"epi": 1,
"er": 28,
"ey": 4,
"f": 41,
"g": 38,
"gcl": 0,
"h#": 12,
"hh": 17,
"hv": 53,
"ih": 2,
"ix": 29,
"iy": 19,
"jh": 59,
"k": 49,
"kcl": 11,
"l": 30,
"m": 40,
"n": 25,
"ng": 52,
"nx": 21,
"ow": 8,
"oy": 39,
"p": 57,
"pau": 14,
"pcl": 16,
"q": 58,
"r": 20,
"s": 43,
"sh": 22,
"t": 33,
"tcl": 23,
"th": 13,
"uh": 7,
"uw": 18,
"ux": 42,
"v": 10,
"w": 35,
"y": 26,
"z": 60,
"zh": 44,
"|": 61
}