Commit
·
329b5c5
1
Parent(s):
24ba6b2
Upload tokenizer
Browse files- added_tokens.json +27 -1
- tokenizer_config.json +0 -1
added_tokens.json
CHANGED
|
@@ -1,4 +1,30 @@
|
|
| 1 |
{
|
| 2 |
"<ctc_blank>": 80,
|
| 3 |
-
"<mask>": 79
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"<ctc_blank>": 80,
|
| 3 |
+
"<mask>": 79,
|
| 4 |
+
"À": 81,
|
| 5 |
+
"Á": 82,
|
| 6 |
+
"È": 83,
|
| 7 |
+
"É": 84,
|
| 8 |
+
"Ì": 85,
|
| 9 |
+
"Í": 86,
|
| 10 |
+
"Ò": 87,
|
| 11 |
+
"Ó": 88,
|
| 12 |
+
"à": 89,
|
| 13 |
+
"á": 90,
|
| 14 |
+
"è": 91,
|
| 15 |
+
"ì": 92,
|
| 16 |
+
"í": 93,
|
| 17 |
+
"ò": 94,
|
| 18 |
+
"ó": 95,
|
| 19 |
+
"ù": 96,
|
| 20 |
+
"ú": 97,
|
| 21 |
+
"ń": 98,
|
| 22 |
+
"Ǹ": 99,
|
| 23 |
+
"ǹ": 100,
|
| 24 |
+
"Ṣ": 101,
|
| 25 |
+
"ṣ": 102,
|
| 26 |
+
"Ẹ": 103,
|
| 27 |
+
"ẹ": 104,
|
| 28 |
+
"Ọ": 105,
|
| 29 |
+
"ọ": 106
|
| 30 |
}
|
tokenizer_config.json
CHANGED
|
@@ -8,6 +8,5 @@
|
|
| 8 |
"processor_class": "SpeechT5Processor",
|
| 9 |
"sp_model_kwargs": {},
|
| 10 |
"tokenizer_class": "SpeechT5Tokenizer",
|
| 11 |
-
"trust_remote_code": false,
|
| 12 |
"unk_token": "<unk>"
|
| 13 |
}
|
|
|
|
| 8 |
"processor_class": "SpeechT5Processor",
|
| 9 |
"sp_model_kwargs": {},
|
| 10 |
"tokenizer_class": "SpeechT5Tokenizer",
|
|
|
|
| 11 |
"unk_token": "<unk>"
|
| 12 |
}
|