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