Upload tokenizer
Browse files- vocab.json +61 -61
vocab.json
CHANGED
@@ -1,66 +1,66 @@
|
|
1 |
{
|
2 |
"[PAD]": 63,
|
3 |
"[UNK]": 62,
|
4 |
-
"aa":
|
5 |
-
"ae":
|
6 |
-
"ah":
|
7 |
-
"ao":
|
8 |
-
"aw":
|
9 |
-
"ax":
|
10 |
-
"ax-h":
|
11 |
-
"axr":
|
12 |
-
"ay":
|
13 |
-
"b":
|
14 |
-
"bcl":
|
15 |
-
"ch":
|
16 |
-
"d":
|
17 |
-
"dcl":
|
18 |
-
"dh":
|
19 |
-
"dx":
|
20 |
-
"eh":
|
21 |
-
"el":
|
22 |
-
"em":
|
23 |
-
"en":
|
24 |
-
"eng":
|
25 |
-
"epi":
|
26 |
-
"er":
|
27 |
-
"ey":
|
28 |
-
"f":
|
29 |
-
"g":
|
30 |
-
"gcl":
|
31 |
-
"h#":
|
32 |
-
"hh":
|
33 |
-
"hv":
|
34 |
-
"ih":
|
35 |
-
"ix":
|
36 |
-
"iy":
|
37 |
-
"jh":
|
38 |
-
"k":
|
39 |
-
"kcl":
|
40 |
-
"l":
|
41 |
-
"m":
|
42 |
-
"n":
|
43 |
-
"ng":
|
44 |
-
"nx":
|
45 |
-
"ow":
|
46 |
-
"oy":
|
47 |
-
"p":
|
48 |
-
"pau":
|
49 |
-
"pcl":
|
50 |
-
"q":
|
51 |
-
"r":
|
52 |
-
"s":
|
53 |
-
"sh":
|
54 |
-
"t":
|
55 |
-
"tcl":
|
56 |
-
"th":
|
57 |
-
"uh":
|
58 |
-
"uw":
|
59 |
-
"ux":
|
60 |
-
"v":
|
61 |
-
"w":
|
62 |
-
"y":
|
63 |
-
"z":
|
64 |
-
"zh":
|
65 |
"|": 61
|
66 |
}
|
|
|
1 |
{
|
2 |
"[PAD]": 63,
|
3 |
"[UNK]": 62,
|
4 |
+
"aa": 37,
|
5 |
+
"ae": 38,
|
6 |
+
"ah": 16,
|
7 |
+
"ao": 9,
|
8 |
+
"aw": 34,
|
9 |
+
"ax": 49,
|
10 |
+
"ax-h": 15,
|
11 |
+
"axr": 5,
|
12 |
+
"ay": 27,
|
13 |
+
"b": 4,
|
14 |
+
"bcl": 1,
|
15 |
+
"ch": 45,
|
16 |
+
"d": 31,
|
17 |
+
"dcl": 18,
|
18 |
+
"dh": 8,
|
19 |
+
"dx": 19,
|
20 |
+
"eh": 12,
|
21 |
+
"el": 44,
|
22 |
+
"em": 22,
|
23 |
+
"en": 7,
|
24 |
+
"eng": 33,
|
25 |
+
"epi": 57,
|
26 |
+
"er": 13,
|
27 |
+
"ey": 20,
|
28 |
+
"f": 29,
|
29 |
+
"g": 59,
|
30 |
+
"gcl": 0,
|
31 |
+
"h#": 21,
|
32 |
+
"hh": 25,
|
33 |
+
"hv": 60,
|
34 |
+
"ih": 42,
|
35 |
+
"ix": 41,
|
36 |
+
"iy": 53,
|
37 |
+
"jh": 51,
|
38 |
+
"k": 40,
|
39 |
+
"kcl": 43,
|
40 |
+
"l": 3,
|
41 |
+
"m": 26,
|
42 |
+
"n": 52,
|
43 |
+
"ng": 39,
|
44 |
+
"nx": 6,
|
45 |
+
"ow": 35,
|
46 |
+
"oy": 2,
|
47 |
+
"p": 56,
|
48 |
+
"pau": 14,
|
49 |
+
"pcl": 48,
|
50 |
+
"q": 58,
|
51 |
+
"r": 30,
|
52 |
+
"s": 55,
|
53 |
+
"sh": 10,
|
54 |
+
"t": 54,
|
55 |
+
"tcl": 47,
|
56 |
+
"th": 36,
|
57 |
+
"uh": 24,
|
58 |
+
"uw": 46,
|
59 |
+
"ux": 11,
|
60 |
+
"v": 17,
|
61 |
+
"w": 23,
|
62 |
+
"y": 32,
|
63 |
+
"z": 50,
|
64 |
+
"zh": 28,
|
65 |
"|": 61
|
66 |
}
|