Commit
•
69f834d
1
Parent(s):
b3b171a
Upload tokenizer
Browse files- added_tokens.json +1 -2
- vocab.json +4 -4
added_tokens.json
CHANGED
|
@@ -1,5 +1,4 @@
|
|
| 1 |
{
|
| 2 |
"</s>": 33,
|
| 3 |
-
"<s>": 32
|
| 4 |
-
"[UNK]": 31
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"</s>": 33,
|
| 3 |
+
"<s>": 32
|
|
|
|
| 4 |
}
|
vocab.json
CHANGED
|
@@ -5,8 +5,10 @@
|
|
| 5 |
"2": 33,
|
| 6 |
"3": 34,
|
| 7 |
"5": 5,
|
|
|
|
| 8 |
"[PAD]": 0,
|
| 9 |
"[UNK]": 31,
|
|
|
|
| 10 |
"b": 10,
|
| 11 |
"c": 35,
|
| 12 |
"d": 16,
|
|
@@ -16,19 +18,17 @@
|
|
| 16 |
"i": 3,
|
| 17 |
"j": 22,
|
| 18 |
"k": 21,
|
| 19 |
-
"
|
| 20 |
"n": 23,
|
| 21 |
"o": 13,
|
| 22 |
"p": 11,
|
| 23 |
"q": 7,
|
|
|
|
| 24 |
"s": 29,
|
| 25 |
"t": 9,
|
| 26 |
"u": 18,
|
| 27 |
-
"v": 19,
|
| 28 |
"w": 24,
|
| 29 |
-
"x": 27,
|
| 30 |
"y": 26,
|
| 31 |
"z": 30,
|
| 32 |
-
"|": 25,
|
| 33 |
"£": 1
|
| 34 |
}
|
|
|
|
| 5 |
"2": 33,
|
| 6 |
"3": 34,
|
| 7 |
"5": 5,
|
| 8 |
+
"9": 8,
|
| 9 |
"[PAD]": 0,
|
| 10 |
"[UNK]": 31,
|
| 11 |
+
"a": 15,
|
| 12 |
"b": 10,
|
| 13 |
"c": 35,
|
| 14 |
"d": 16,
|
|
|
|
| 18 |
"i": 3,
|
| 19 |
"j": 22,
|
| 20 |
"k": 21,
|
| 21 |
+
"l": 28,
|
| 22 |
"n": 23,
|
| 23 |
"o": 13,
|
| 24 |
"p": 11,
|
| 25 |
"q": 7,
|
| 26 |
+
"r": 17,
|
| 27 |
"s": 29,
|
| 28 |
"t": 9,
|
| 29 |
"u": 18,
|
|
|
|
| 30 |
"w": 24,
|
|
|
|
| 31 |
"y": 26,
|
| 32 |
"z": 30,
|
|
|
|
| 33 |
"£": 1
|
| 34 |
}
|