Upload tokenizer
Browse files- added_tokens.json +2 -2
- special_tokens_map.json +1 -1
- tokenizer_config.json +1 -1
- vocab.json +1 -2
added_tokens.json
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
{
|
| 2 |
-
"</s>":
|
| 3 |
-
"<s>":
|
| 4 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"</s>": 29,
|
| 3 |
+
"<s>": 28
|
| 4 |
}
|
special_tokens_map.json
CHANGED
|
@@ -18,5 +18,5 @@
|
|
| 18 |
"bos_token": "<s>",
|
| 19 |
"eos_token": "</s>",
|
| 20 |
"pad_token": "[PAD]",
|
| 21 |
-
"unk_token": "
|
| 22 |
}
|
|
|
|
| 18 |
"bos_token": "<s>",
|
| 19 |
"eos_token": "</s>",
|
| 20 |
"pad_token": "[PAD]",
|
| 21 |
+
"unk_token": "<unk>"
|
| 22 |
}
|
tokenizer_config.json
CHANGED
|
@@ -7,6 +7,6 @@
|
|
| 7 |
"replace_word_delimiter_char": " ",
|
| 8 |
"special_tokens_map_file": null,
|
| 9 |
"tokenizer_class": "Wav2Vec2CTCTokenizer",
|
| 10 |
-
"unk_token": "
|
| 11 |
"word_delimiter_token": "|"
|
| 12 |
}
|
|
|
|
| 7 |
"replace_word_delimiter_char": " ",
|
| 8 |
"special_tokens_map_file": null,
|
| 9 |
"tokenizer_class": "Wav2Vec2CTCTokenizer",
|
| 10 |
+
"unk_token": "<unk>",
|
| 11 |
"word_delimiter_token": "|"
|
| 12 |
}
|
vocab.json
CHANGED
|
@@ -1,6 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"[PAD]":
|
| 3 |
-
"[UNK]": 27,
|
| 4 |
"a": 1,
|
| 5 |
"b": 2,
|
| 6 |
"c": 3,
|
|
|
|
| 1 |
{
|
| 2 |
+
"[PAD]": 27,
|
|
|
|
| 3 |
"a": 1,
|
| 4 |
"b": 2,
|
| 5 |
"c": 3,
|