aaniket's picture
Upload tokenizer
3f2ce83
raw history blame
No virus
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 46,
"ae": 8,
"ah": 50,
"ao": 39,
"aw": 45,
"ax": 18,
"ax-h": 42,
"axr": 7,
"ay": 52,
"b": 10,
"bcl": 29,
"ch": 1,
"d": 12,
"dcl": 0,
"dh": 27,
"dx": 23,
"eh": 53,
"el": 58,
"em": 60,
"en": 22,
"eng": 20,
"epi": 49,
"er": 34,
"ey": 19,
"f": 38,
"g": 59,
"gcl": 37,
"h#": 55,
"hh": 41,
"hv": 15,
"ih": 13,
"ix": 43,
"iy": 44,
"jh": 9,
"k": 40,
"kcl": 31,
"l": 36,
"m": 17,
"n": 5,
"ng": 21,
"nx": 24,
"ow": 28,
"oy": 48,
"p": 32,
"pau": 33,
"pcl": 4,
"q": 30,
"r": 51,
"s": 54,
"sh": 11,
"t": 14,
"tcl": 47,
"th": 6,
"uh": 3,
"uw": 2,
"ux": 56,
"v": 35,
"w": 16,
"y": 25,
"z": 26,
"zh": 57,
"|": 61
}