Fix tokenizer_config.json: use PreTrainedTokenizerFast (bbert_768H_epoch100)
Browse files
bbert_768H_epoch100/tokenizer_config.json
CHANGED
|
@@ -1,11 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"backend": "tokenizers",
|
| 3 |
"bos_token": "<s>",
|
| 4 |
"clean_up_tokenization_spaces": true,
|
| 5 |
"cls_token": "<cls>",
|
| 6 |
"eos_token": "</s>",
|
| 7 |
-
"is_local": true,
|
| 8 |
-
"local_files_only": false,
|
| 9 |
"mask_token": "<msk>",
|
| 10 |
"max_length": 102,
|
| 11 |
"model_max_length": 512,
|
|
@@ -15,8 +12,8 @@
|
|
| 15 |
"padding_side": "right",
|
| 16 |
"sep_token": "<sep>",
|
| 17 |
"stride": 0,
|
| 18 |
-
"tokenizer_class": "
|
| 19 |
"truncation_side": "right",
|
| 20 |
"truncation_strategy": "longest_first",
|
| 21 |
"unk_token": "<unk>"
|
| 22 |
-
}
|
|
|
|
| 1 |
{
|
|
|
|
| 2 |
"bos_token": "<s>",
|
| 3 |
"clean_up_tokenization_spaces": true,
|
| 4 |
"cls_token": "<cls>",
|
| 5 |
"eos_token": "</s>",
|
|
|
|
|
|
|
| 6 |
"mask_token": "<msk>",
|
| 7 |
"max_length": 102,
|
| 8 |
"model_max_length": 512,
|
|
|
|
| 12 |
"padding_side": "right",
|
| 13 |
"sep_token": "<sep>",
|
| 14 |
"stride": 0,
|
| 15 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 16 |
"truncation_side": "right",
|
| 17 |
"truncation_strategy": "longest_first",
|
| 18 |
"unk_token": "<unk>"
|
| 19 |
+
}
|