LeoCordoba
commited on
Commit
•
71b110c
1
Parent(s):
fec8fca
Update tokenizer_config.json
Browse files- tokenizer_config.json +13 -1
tokenizer_config.json
CHANGED
@@ -1 +1,13 @@
|
|
1 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_lower_case": true,
|
3 |
+
"unk_token": "[UNK]",
|
4 |
+
"sep_token": "[SEP]",
|
5 |
+
"pad_token": "[PAD]",
|
6 |
+
"cls_token": "[CLS]",
|
7 |
+
"mask_token": "[MASK]",
|
8 |
+
"tokenize_chinese_chars": true,
|
9 |
+
"strip_accents": false,
|
10 |
+
"do_basic_tokenize": true,
|
11 |
+
"never_split": null,
|
12 |
+
"model_max_length": 512
|
13 |
+
}
|