Commit
•
c289364
1
Parent(s):
8e3929c
Update tokenizer_config.json (#4)
Browse files- Update tokenizer_config.json (d31b7625019e06d9904d3210e38fbfad52fc4e7f)
Co-authored-by: Raj Sanjay Shah <Raj-Sanjay-Shah@users.noreply.huggingface.co>
- tokenizer_config.json +9 -2
tokenizer_config.json
CHANGED
|
@@ -43,14 +43,21 @@
|
|
| 43 |
"special": true
|
| 44 |
}
|
| 45 |
},
|
| 46 |
-
"bos_token":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
"clean_up_tokenization_spaces": true,
|
| 48 |
"cls_token": "<s>",
|
| 49 |
"eos_token": "</s>",
|
| 50 |
"errors": "replace",
|
| 51 |
"mask_token": "<mask>",
|
| 52 |
"max_length": 128,
|
| 53 |
-
"model_max_length":
|
| 54 |
"pad_to_multiple_of": null,
|
| 55 |
"pad_token": "<pad>",
|
| 56 |
"pad_token_type_id": 0,
|
|
|
|
| 43 |
"special": true
|
| 44 |
}
|
| 45 |
},
|
| 46 |
+
"bos_token": {
|
| 47 |
+
"content": "<s>",
|
| 48 |
+
"single_word": false,
|
| 49 |
+
"lstrip": false,
|
| 50 |
+
"rstrip": false,
|
| 51 |
+
"normalized": true,
|
| 52 |
+
"__type": "AddedToken"
|
| 53 |
+
},
|
| 54 |
"clean_up_tokenization_spaces": true,
|
| 55 |
"cls_token": "<s>",
|
| 56 |
"eos_token": "</s>",
|
| 57 |
"errors": "replace",
|
| 58 |
"mask_token": "<mask>",
|
| 59 |
"max_length": 128,
|
| 60 |
+
"model_max_length": 1e+30,
|
| 61 |
"pad_to_multiple_of": null,
|
| 62 |
"pad_token": "<pad>",
|
| 63 |
"pad_token_type_id": 0,
|