aaniket's picture
Upload tokenizer
21076d7
raw
history blame contribute delete
750 Bytes
{
"[PAD]": 62,
"[UNK]": 61,
"aa": 9,
"ae": 54,
"ah": 17,
"ao": 18,
"aw": 46,
"ax": 5,
"ax-h": 3,
"axr": 1,
"ay": 48,
"b": 19,
"bcl": 32,
"ch": 28,
"d": 34,
"dcl": 37,
"dh": 47,
"dx": 31,
"eh": 22,
"el": 33,
"em": 2,
"en": 36,
"eng": 57,
"epi": 16,
"er": 38,
"ey": 24,
"f": 60,
"g": 30,
"gcl": 23,
"h#": 53,
"hh": 11,
"hv": 58,
"ih": 4,
"ix": 49,
"iy": 42,
"jh": 14,
"k": 10,
"kcl": 41,
"l": 56,
"m": 27,
"n": 44,
"ng": 39,
"nx": 45,
"ow": 15,
"oy": 51,
"p": 6,
"pau": 59,
"pcl": 50,
"q": 0,
"r": 13,
"s": 35,
"sh": 55,
"t": 43,
"tcl": 8,
"th": 52,
"uh": 21,
"uw": 40,
"ux": 25,
"v": 12,
"w": 29,
"y": 26,
"z": 20,
"zh": 7
}