change config
Browse files- tokenizer_config.json +8 -1
tokenizer_config.json
CHANGED
|
@@ -5,7 +5,14 @@
|
|
| 5 |
"sep_token": "[SEP]",
|
| 6 |
"cls_token": "[CLS]",
|
| 7 |
"pad_token": "<pad>",
|
| 8 |
-
"mask_token":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
"model_max_length": 512,
|
| 10 |
"special_tokens_map_file": null,
|
| 11 |
"name_or_path": "SaulLu/bengali-tokenizer-v2"
|
|
|
|
| 5 |
"sep_token": "[SEP]",
|
| 6 |
"cls_token": "[CLS]",
|
| 7 |
"pad_token": "<pad>",
|
| 8 |
+
"mask_token": {
|
| 9 |
+
"content": "[MASK]",
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"lstrip": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"normalized": false,
|
| 14 |
+
"__type": "AddedToken"
|
| 15 |
+
},
|
| 16 |
"model_max_length": 512,
|
| 17 |
"special_tokens_map_file": null,
|
| 18 |
"name_or_path": "SaulLu/bengali-tokenizer-v2"
|