Upload tokenizer
Browse files- vocab.json +114 -114
vocab.json
CHANGED
@@ -1,118 +1,118 @@
|
|
1 |
{
|
2 |
-
"aɪ":
|
3 |
-
"aɪə":
|
4 |
-
"aɪɚ":
|
5 |
-
"aʊ":
|
6 |
-
"b":
|
7 |
-
"d":
|
8 |
-
"dʒ":
|
9 |
-
"eɪ":
|
10 |
-
"f":
|
11 |
-
"h":
|
12 |
-
"i":
|
13 |
-
"iə":
|
14 |
-
"iː":
|
15 |
-
"j":
|
16 |
-
"k":
|
17 |
-
"l":
|
18 |
-
"m":
|
19 |
-
"n":
|
20 |
-
"n̩":
|
21 |
-
"oʊ":
|
22 |
-
"oː":
|
23 |
-
"oːɹ":
|
24 |
-
"p":
|
25 |
-
"r":
|
26 |
-
"s":
|
27 |
-
"t":
|
28 |
-
"tʃ":
|
29 |
-
"uː":
|
30 |
-
"v":
|
31 |
-
"w":
|
32 |
-
"x":
|
33 |
-
"z":
|
34 |
"|": 0,
|
35 |
-
"æ":
|
36 |
-
"ð":
|
37 |
-
"ŋ":
|
38 |
-
"ɐ":
|
39 |
-
"ɑː":
|
40 |
-
"ɑːɹ":
|
41 |
-
"ɔ":
|
42 |
-
"ɔɪ":
|
43 |
-
"ɔː":
|
44 |
-
"ɔːɹ":
|
45 |
-
"ə":
|
46 |
-
"əl":
|
47 |
-
"ɚ":
|
48 |
-
"ɛ":
|
49 |
-
"ɛɹ":
|
50 |
-
"ɜː":
|
51 |
-
"ɡ":
|
52 |
-
"ɪ":
|
53 |
-
"ɪɹ":
|
54 |
-
"ɹ":
|
55 |
-
"ɾ":
|
56 |
-
"ʃ":
|
57 |
-
"ʊ":
|
58 |
-
"ʊɹ":
|
59 |
-
"ʌ":
|
60 |
-
"ʒ":
|
61 |
"ʔ": 66,
|
62 |
-
"ˈaɪ":
|
63 |
-
"ˈaɪə":
|
64 |
-
"ˈaɪɚ":
|
65 |
-
"ˈaʊ":
|
66 |
-
"ˈeɪ":
|
67 |
-
"ˈiə":
|
68 |
-
"ˈiː":
|
69 |
-
"ˈoʊ":
|
70 |
-
"ˈoː":
|
71 |
-
"ˈoːɹ":
|
72 |
-
"ˈuː":
|
73 |
-
"ˈæ":
|
74 |
-
"ˈɑː":
|
75 |
-
"ˈɑːɹ":
|
76 |
-
"ˈɔ":
|
77 |
-
"ˈɔɪ":
|
78 |
-
"ˈɔː":
|
79 |
-
"ˈɔːɹ":
|
80 |
-
"ˈɚ":
|
81 |
-
"ˈɛ":
|
82 |
-
"ˈɛɹ":
|
83 |
-
"ˈɜː":
|
84 |
-
"ˈɪ":
|
85 |
-
"ˈɪɹ":
|
86 |
-
"ˈʊ":
|
87 |
-
"ˈʊɹ":
|
88 |
-
"ˈʌ":
|
89 |
-
"ˌaɪ":
|
90 |
-
"ˌaɪɚ":
|
91 |
-
"ˌaʊ":
|
92 |
-
"ˌeɪ":
|
93 |
-
"ˌiə":
|
94 |
-
"ˌiː":
|
95 |
-
"ˌn̩":
|
96 |
-
"ˌoʊ":
|
97 |
-
"ˌoː":
|
98 |
-
"ˌoːɹ":
|
99 |
-
"ˌuː":
|
100 |
-
"ˌæ":
|
101 |
-
"ˌɐ":
|
102 |
-
"ˌɑː":
|
103 |
-
"ˌɑːɹ":
|
104 |
-
"ˌɔ":
|
105 |
-
"ˌɔɪ":
|
106 |
-
"ˌɔː":
|
107 |
-
"ˌɔːɹ":
|
108 |
-
"ˌɛ":
|
109 |
-
"ˌɛɹ":
|
110 |
-
"ˌɜː":
|
111 |
-
"ˌɪ":
|
112 |
-
"ˌɪɹ":
|
113 |
-
"ˌʊ":
|
114 |
-
"ˌʊɹ":
|
115 |
-
"ˌʌ":
|
116 |
-
"θ":
|
117 |
-
"ᵻ":
|
118 |
}
|
|
|
1 |
{
|
2 |
+
"aɪ": 16,
|
3 |
+
"aɪə": 70,
|
4 |
+
"aɪɚ": 60,
|
5 |
+
"aʊ": 33,
|
6 |
+
"b": 93,
|
7 |
+
"d": 34,
|
8 |
+
"dʒ": 43,
|
9 |
+
"eɪ": 46,
|
10 |
+
"f": 112,
|
11 |
+
"h": 92,
|
12 |
+
"i": 97,
|
13 |
+
"iə": 2,
|
14 |
+
"iː": 27,
|
15 |
+
"j": 88,
|
16 |
+
"k": 32,
|
17 |
+
"l": 74,
|
18 |
+
"m": 99,
|
19 |
+
"n": 7,
|
20 |
+
"n̩": 25,
|
21 |
+
"oʊ": 20,
|
22 |
+
"oː": 9,
|
23 |
+
"oːɹ": 94,
|
24 |
+
"p": 73,
|
25 |
+
"r": 18,
|
26 |
+
"s": 55,
|
27 |
+
"t": 51,
|
28 |
+
"tʃ": 96,
|
29 |
+
"uː": 52,
|
30 |
+
"v": 19,
|
31 |
+
"w": 76,
|
32 |
+
"x": 71,
|
33 |
+
"z": 21,
|
34 |
"|": 0,
|
35 |
+
"æ": 26,
|
36 |
+
"ð": 79,
|
37 |
+
"ŋ": 61,
|
38 |
+
"ɐ": 104,
|
39 |
+
"ɑː": 103,
|
40 |
+
"ɑːɹ": 31,
|
41 |
+
"ɔ": 114,
|
42 |
+
"ɔɪ": 24,
|
43 |
+
"ɔː": 8,
|
44 |
+
"ɔːɹ": 44,
|
45 |
+
"ə": 35,
|
46 |
+
"əl": 90,
|
47 |
+
"ɚ": 110,
|
48 |
+
"ɛ": 82,
|
49 |
+
"ɛɹ": 29,
|
50 |
+
"ɜː": 65,
|
51 |
+
"ɡ": 6,
|
52 |
+
"ɪ": 14,
|
53 |
+
"ɪɹ": 62,
|
54 |
+
"ɹ": 83,
|
55 |
+
"ɾ": 11,
|
56 |
+
"ʃ": 4,
|
57 |
+
"ʊ": 15,
|
58 |
+
"ʊɹ": 78,
|
59 |
+
"ʌ": 1,
|
60 |
+
"ʒ": 91,
|
61 |
"ʔ": 66,
|
62 |
+
"ˈaɪ": 45,
|
63 |
+
"ˈaɪə": 22,
|
64 |
+
"ˈaɪɚ": 40,
|
65 |
+
"ˈaʊ": 86,
|
66 |
+
"ˈeɪ": 53,
|
67 |
+
"ˈiə": 77,
|
68 |
+
"ˈiː": 3,
|
69 |
+
"ˈoʊ": 85,
|
70 |
+
"ˈoː": 109,
|
71 |
+
"ˈoːɹ": 13,
|
72 |
+
"ˈuː": 115,
|
73 |
+
"ˈæ": 57,
|
74 |
+
"ˈɑː": 102,
|
75 |
+
"ˈɑːɹ": 87,
|
76 |
+
"ˈɔ": 41,
|
77 |
+
"ˈɔɪ": 38,
|
78 |
+
"ˈɔː": 36,
|
79 |
+
"ˈɔːɹ": 68,
|
80 |
+
"ˈɚ": 64,
|
81 |
+
"ˈɛ": 72,
|
82 |
+
"ˈɛɹ": 63,
|
83 |
+
"ˈɜː": 37,
|
84 |
+
"ˈɪ": 59,
|
85 |
+
"ˈɪɹ": 108,
|
86 |
+
"ˈʊ": 80,
|
87 |
+
"ˈʊɹ": 17,
|
88 |
+
"ˈʌ": 42,
|
89 |
+
"ˌaɪ": 56,
|
90 |
+
"ˌaɪɚ": 50,
|
91 |
+
"ˌaʊ": 23,
|
92 |
+
"ˌeɪ": 47,
|
93 |
+
"ˌiə": 12,
|
94 |
+
"ˌiː": 98,
|
95 |
+
"ˌn̩": 10,
|
96 |
+
"ˌoʊ": 54,
|
97 |
+
"ˌoː": 67,
|
98 |
+
"ˌoːɹ": 30,
|
99 |
+
"ˌuː": 39,
|
100 |
+
"ˌæ": 58,
|
101 |
+
"ˌɐ": 48,
|
102 |
+
"ˌɑː": 28,
|
103 |
+
"ˌɑːɹ": 113,
|
104 |
+
"ˌɔ": 95,
|
105 |
+
"ˌɔɪ": 100,
|
106 |
+
"ˌɔː": 75,
|
107 |
+
"ˌɔːɹ": 111,
|
108 |
+
"ˌɛ": 101,
|
109 |
+
"ˌɛɹ": 106,
|
110 |
+
"ˌɜː": 84,
|
111 |
+
"ˌɪ": 5,
|
112 |
+
"ˌɪɹ": 81,
|
113 |
+
"ˌʊ": 49,
|
114 |
+
"ˌʊɹ": 105,
|
115 |
+
"ˌʌ": 69,
|
116 |
+
"θ": 89,
|
117 |
+
"ᵻ": 107
|
118 |
}
|