Upload tokenizer
Browse files- special_tokens_map.json +7 -1
- tokenizer_config.json +6 -1
special_tokens_map.json
CHANGED
|
@@ -1 +1,7 @@
|
|
| 1 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"mask_token": "[MASK]",
|
| 4 |
+
"pad_token": "[PAD]",
|
| 5 |
+
"sep_token": "[SEP]",
|
| 6 |
+
"unk_token": "[UNK]"
|
| 7 |
+
}
|
tokenizer_config.json
CHANGED
|
@@ -6170,6 +6170,11 @@
|
|
| 6170 |
}
|
| 6171 |
},
|
| 6172 |
"clean_up_tokenization_spaces": true,
|
|
|
|
|
|
|
| 6173 |
"model_max_length": 1000000000000000019884624838656,
|
| 6174 |
-
"
|
|
|
|
|
|
|
|
|
|
| 6175 |
}
|
|
|
|
| 6170 |
}
|
| 6171 |
},
|
| 6172 |
"clean_up_tokenization_spaces": true,
|
| 6173 |
+
"cls_token": "[CLS]",
|
| 6174 |
+
"mask_token": "[MASK]",
|
| 6175 |
"model_max_length": 1000000000000000019884624838656,
|
| 6176 |
+
"pad_token": "[PAD]",
|
| 6177 |
+
"sep_token": "[SEP]",
|
| 6178 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 6179 |
+
"unk_token": "[UNK]"
|
| 6180 |
}
|