Upload 2 files
Browse files- tokenizer.json +0 -0
- tokenizer_config.json +32 -0
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<special0>",
|
| 4 |
+
"<special1>",
|
| 5 |
+
"<special2>",
|
| 6 |
+
"<special3>",
|
| 7 |
+
"<special4>",
|
| 8 |
+
"<special5>",
|
| 9 |
+
"<special6>",
|
| 10 |
+
"<special7>",
|
| 11 |
+
"<special8>",
|
| 12 |
+
"<special9>"
|
| 13 |
+
],
|
| 14 |
+
"bos_token": "<s>",
|
| 15 |
+
"clean_up_tokenization_spaces": true,
|
| 16 |
+
"cls_token": "</s>",
|
| 17 |
+
"do_lowercase_and_remove_accent": true,
|
| 18 |
+
"id2lang": {
|
| 19 |
+
"0": "de",
|
| 20 |
+
"1": "en"
|
| 21 |
+
},
|
| 22 |
+
"lang2id": {
|
| 23 |
+
"de": 0,
|
| 24 |
+
"en": 1
|
| 25 |
+
},
|
| 26 |
+
"mask_token": "<special1>",
|
| 27 |
+
"model_max_length": 512,
|
| 28 |
+
"pad_token": "<pad>",
|
| 29 |
+
"sep_token": "</s>",
|
| 30 |
+
"tokenizer_class": "XLMTokenizer",
|
| 31 |
+
"unk_token": "<unk>"
|
| 32 |
+
}
|