Update tokenizer_config.json
Browse files- tokenizer_config.json +8 -1
tokenizer_config.json
CHANGED
|
@@ -33,7 +33,14 @@
|
|
| 33 |
"special": true
|
| 34 |
}
|
| 35 |
},
|
| 36 |
-
"pre_tokenizer":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
"bos_token": "<s>",
|
| 38 |
"clean_up_tokenization_spaces": true,
|
| 39 |
"do_lower_case": false,
|
|
|
|
| 33 |
"special": true
|
| 34 |
}
|
| 35 |
},
|
| 36 |
+
"pre_tokenizer": {
|
| 37 |
+
"type": "Split",
|
| 38 |
+
"pattern": {
|
| 39 |
+
"Regex": ""
|
| 40 |
+
},
|
| 41 |
+
"behavior": "Isolated",
|
| 42 |
+
"invert": false
|
| 43 |
+
},
|
| 44 |
"bos_token": "<s>",
|
| 45 |
"clean_up_tokenization_spaces": true,
|
| 46 |
"do_lower_case": false,
|