Commit
•
4d31b35
1
Parent(s):
246f8ef
Upload tokenizer
Browse files- special_tokens_map.json +7 -1
- tokenizer_config.json +4 -0
special_tokens_map.json
CHANGED
|
@@ -19,7 +19,13 @@
|
|
| 19 |
"rstrip": false,
|
| 20 |
"single_word": false
|
| 21 |
},
|
| 22 |
-
"mask_token":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
"pad_token": {
|
| 24 |
"content": "<pad>",
|
| 25 |
"lstrip": false,
|
|
|
|
| 19 |
"rstrip": false,
|
| 20 |
"single_word": false
|
| 21 |
},
|
| 22 |
+
"mask_token": {
|
| 23 |
+
"content": "_",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false
|
| 28 |
+
},
|
| 29 |
"pad_token": {
|
| 30 |
"content": "<pad>",
|
| 31 |
"lstrip": false,
|
tokenizer_config.json
CHANGED
|
@@ -51,8 +51,12 @@
|
|
| 51 |
"clean_up_tokenization_spaces": true,
|
| 52 |
"eos_token": "</s>",
|
| 53 |
"mask_token": "_",
|
|
|
|
| 54 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
|
| 55 |
"pad_token": "<pad>",
|
|
|
|
|
|
|
| 56 |
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 57 |
"unk_token": "<unk>"
|
| 58 |
}
|
|
|
|
| 51 |
"clean_up_tokenization_spaces": true,
|
| 52 |
"eos_token": "</s>",
|
| 53 |
"mask_token": "_",
|
| 54 |
+
"max_length": null,
|
| 55 |
"model_max_length": 1000000000000000019884624838656,
|
| 56 |
+
"pad_to_multiple_of": null,
|
| 57 |
"pad_token": "<pad>",
|
| 58 |
+
"pad_token_type_id": 0,
|
| 59 |
+
"padding_side": "right",
|
| 60 |
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 61 |
"unk_token": "<unk>"
|
| 62 |
}
|