Upload bert_tokenizer.py
Browse files- bert_tokenizer.py +4 -4
bert_tokenizer.py
CHANGED
@@ -18,10 +18,10 @@ from transformers import BertTokenizerFast
|
|
18 |
cache_path = Path(os.path.abspath(__file__)).parent
|
19 |
|
20 |
SOURCE_FILES_URL = {
|
21 |
-
"vocab.txt": "https://huggingface.co/iioSnail/chinesebert-base/
|
22 |
-
"pinyin_map.json": "https://huggingface.co/iioSnail/chinesebert-base/
|
23 |
-
"id2pinyin.json": "https://huggingface.co/iioSnail/chinesebert-base/
|
24 |
-
"pinyin2tensor.json": "https://huggingface.co/iioSnail/chinesebert-base/
|
25 |
}
|
26 |
|
27 |
|
|
|
18 |
cache_path = Path(os.path.abspath(__file__)).parent
|
19 |
|
20 |
SOURCE_FILES_URL = {
|
21 |
+
"vocab.txt": "https://huggingface.co/iioSnail/chinesebert-base/resolve/main/vocab.txt",
|
22 |
+
"pinyin_map.json": "https://huggingface.co/iioSnail/chinesebert-base/resolve/main/config/pinyin_map.json",
|
23 |
+
"id2pinyin.json": "https://huggingface.co/iioSnail/chinesebert-base/resolve/main/config/id2pinyin.json",
|
24 |
+
"pinyin2tensor.json": "https://huggingface.co/iioSnail/chinesebert-base/resolve/main/config/id2pinyin.json",
|
25 |
}
|
26 |
|
27 |
|