aaniket's picture
Upload tokenizer
aede03a
raw
history blame
761 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"aa": 0,
"ae": 9,
"ah": 25,
"ao": 8,
"aw": 21,
"ax": 15,
"ax-h": 1,
"axr": 53,
"ay": 41,
"b": 14,
"bcl": 60,
"ch": 28,
"d": 17,
"dcl": 12,
"dh": 51,
"dx": 56,
"eh": 4,
"el": 31,
"em": 49,
"en": 46,
"eng": 20,
"epi": 48,
"er": 24,
"ey": 57,
"f": 37,
"g": 38,
"gcl": 32,
"h#": 19,
"hh": 55,
"hv": 29,
"ih": 35,
"ix": 40,
"iy": 33,
"jh": 5,
"k": 13,
"kcl": 23,
"l": 36,
"m": 44,
"n": 30,
"ng": 50,
"nx": 27,
"ow": 59,
"oy": 54,
"p": 10,
"pau": 47,
"pcl": 45,
"q": 42,
"r": 39,
"s": 22,
"sh": 11,
"t": 16,
"tcl": 2,
"th": 18,
"uh": 7,
"uw": 52,
"ux": 6,
"v": 26,
"w": 34,
"y": 43,
"z": 3,
"zh": 58,
"|": 61
}