aaniket's picture
Upload tokenizer
9d80e28
raw
history blame
No virus
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 16,
"ae": 46,
"ah": 25,
"ao": 50,
"aw": 22,
"ax": 59,
"ax-h": 24,
"axr": 23,
"ay": 0,
"b": 29,
"bcl": 34,
"ch": 27,
"d": 12,
"dcl": 58,
"dh": 49,
"dx": 45,
"eh": 8,
"el": 57,
"em": 10,
"en": 4,
"eng": 52,
"epi": 35,
"er": 56,
"ey": 41,
"f": 47,
"g": 33,
"gcl": 38,
"h#": 55,
"hh": 51,
"hv": 36,
"ih": 21,
"ix": 17,
"iy": 37,
"jh": 39,
"k": 19,
"kcl": 11,
"l": 44,
"m": 48,
"n": 60,
"ng": 28,
"nx": 7,
"ow": 3,
"oy": 9,
"p": 32,
"pau": 18,
"pcl": 14,
"q": 43,
"r": 31,
"s": 1,
"sh": 40,
"t": 53,
"tcl": 54,
"th": 20,
"uh": 15,
"uw": 26,
"ux": 2,
"v": 42,
"w": 13,
"y": 6,
"z": 5,
"zh": 30,
"|": 61
}