C. Servan commited on
Commit ·
e51c021
1
Parent(s): ffe3a2b
update tokenizer config
Browse files- special_tokens_map.json +2 -0
- tokenizer_config.json +2 -0
special_tokens_map.json
CHANGED
|
@@ -1,7 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"cls_token": "[CLS]",
|
|
|
|
| 3 |
"mask_token": "[MASK]",
|
| 4 |
"pad_token": "[PAD]",
|
| 5 |
"sep_token": "[SEP]",
|
|
|
|
| 6 |
"unk_token": "[UNK]"
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"cls_token": "[CLS]",
|
| 3 |
+
"bos_token": "[CLS]",
|
| 4 |
"mask_token": "[MASK]",
|
| 5 |
"pad_token": "[PAD]",
|
| 6 |
"sep_token": "[SEP]",
|
| 7 |
+
"eos_token": "[SEP]",
|
| 8 |
"unk_token": "[UNK]"
|
| 9 |
}
|
tokenizer_config.json
CHANGED
|
@@ -8043,6 +8043,7 @@
|
|
| 8043 |
},
|
| 8044 |
"clean_up_tokenization_spaces": true,
|
| 8045 |
"cls_token": "[CLS]",
|
|
|
|
| 8046 |
"extra_special_tokens": {},
|
| 8047 |
"mask_token": "[MASK]",
|
| 8048 |
"model_input_names": [
|
|
@@ -8052,6 +8053,7 @@
|
|
| 8052 |
"model_max_length": 8192,
|
| 8053 |
"pad_token": "[PAD]",
|
| 8054 |
"sep_token": "[SEP]",
|
|
|
|
| 8055 |
"tokenizer_class": "PreTrainedTokenizer",
|
| 8056 |
"unk_token": "[UNK]"
|
| 8057 |
}
|
|
|
|
| 8043 |
},
|
| 8044 |
"clean_up_tokenization_spaces": true,
|
| 8045 |
"cls_token": "[CLS]",
|
| 8046 |
+
"bos_token": "[CLS]",
|
| 8047 |
"extra_special_tokens": {},
|
| 8048 |
"mask_token": "[MASK]",
|
| 8049 |
"model_input_names": [
|
|
|
|
| 8053 |
"model_max_length": 8192,
|
| 8054 |
"pad_token": "[PAD]",
|
| 8055 |
"sep_token": "[SEP]",
|
| 8056 |
+
"eos_token": "[SEP]",
|
| 8057 |
"tokenizer_class": "PreTrainedTokenizer",
|
| 8058 |
"unk_token": "[UNK]"
|
| 8059 |
}
|