Upload tokenizer
Browse files- vocab.json +40 -40
vocab.json
CHANGED
@@ -1,48 +1,48 @@
|
|
1 |
{
|
2 |
"[PAD]": 45,
|
3 |
"[UNK]": 44,
|
4 |
-
"a":
|
5 |
-
"b":
|
6 |
-
"d":
|
7 |
-
"e":
|
8 |
-
"f":
|
9 |
-
"h":
|
10 |
-
"i":
|
11 |
-
"j":
|
12 |
-
"k":
|
13 |
-
"l":
|
14 |
-
"m":
|
15 |
-
"n":
|
16 |
-
"o":
|
17 |
-
"p":
|
18 |
-
"s":
|
19 |
"t": 37,
|
20 |
"u": 33,
|
21 |
-
"v":
|
22 |
-
"w":
|
23 |
-
"z":
|
24 |
-
"|":
|
25 |
-
"æ":
|
26 |
-
"ð":
|
27 |
"ŋ": 28,
|
28 |
-
"ɐ":
|
29 |
-
"ɑ":
|
30 |
-
"ɔ":
|
31 |
-
"ə":
|
32 |
-
"ɚ":
|
33 |
-
"ɛ":
|
34 |
-
"ɜ":
|
35 |
-
"ɡ":
|
36 |
-
"ɪ":
|
37 |
-
"ɹ":
|
38 |
-
"ɾ":
|
39 |
-
"ʃ":
|
40 |
-
"ʊ":
|
41 |
-
"ʌ":
|
42 |
"ʒ": 15,
|
43 |
-
"ʔ":
|
44 |
-
"ː":
|
45 |
-
"̩":
|
46 |
-
"θ":
|
47 |
-
"ᵻ":
|
48 |
}
|
|
|
1 |
{
|
2 |
"[PAD]": 45,
|
3 |
"[UNK]": 44,
|
4 |
+
"a": 16,
|
5 |
+
"b": 31,
|
6 |
+
"d": 23,
|
7 |
+
"e": 14,
|
8 |
+
"f": 27,
|
9 |
+
"h": 19,
|
10 |
+
"i": 0,
|
11 |
+
"j": 30,
|
12 |
+
"k": 12,
|
13 |
+
"l": 7,
|
14 |
+
"m": 42,
|
15 |
+
"n": 43,
|
16 |
+
"o": 38,
|
17 |
+
"p": 32,
|
18 |
+
"s": 3,
|
19 |
"t": 37,
|
20 |
"u": 33,
|
21 |
+
"v": 1,
|
22 |
+
"w": 40,
|
23 |
+
"z": 39,
|
24 |
+
"|": 4,
|
25 |
+
"æ": 18,
|
26 |
+
"ð": 6,
|
27 |
"ŋ": 28,
|
28 |
+
"ɐ": 9,
|
29 |
+
"ɑ": 2,
|
30 |
+
"ɔ": 41,
|
31 |
+
"ə": 8,
|
32 |
+
"ɚ": 13,
|
33 |
+
"ɛ": 22,
|
34 |
+
"ɜ": 24,
|
35 |
+
"ɡ": 10,
|
36 |
+
"ɪ": 34,
|
37 |
+
"ɹ": 5,
|
38 |
+
"ɾ": 26,
|
39 |
+
"ʃ": 29,
|
40 |
+
"ʊ": 36,
|
41 |
+
"ʌ": 21,
|
42 |
"ʒ": 15,
|
43 |
+
"ʔ": 11,
|
44 |
+
"ː": 17,
|
45 |
+
"̩": 25,
|
46 |
+
"θ": 35,
|
47 |
+
"ᵻ": 20
|
48 |
}
|