Commit
•
ad15617
1
Parent(s):
ecca05a
Update tokenizer_config.json
Browse files- tokenizer_config.json +10 -4
tokenizer_config.json
CHANGED
|
@@ -2,8 +2,14 @@
|
|
| 2 |
"auto_map": {
|
| 3 |
"AutoTokenizer": ["tokenization_bn.BNTokenizer", null]
|
| 4 |
},
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
| 8 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"auto_map": {
|
| 3 |
"AutoTokenizer": ["tokenization_bn.BNTokenizer", null]
|
| 4 |
},
|
| 5 |
+
"tokenizer_class": "BNTokenizer",
|
| 6 |
+
"bos_token": null,
|
| 7 |
+
"clean_up_tokenization_spaces": false,
|
| 8 |
+
"eos_token": "</s>",
|
| 9 |
+
"model_max_length": 2048,
|
| 10 |
+
"pad_token": "<|reserved001|>",
|
| 11 |
+
"padding_side": "right",
|
| 12 |
+
"sep_token": null,
|
| 13 |
+
"sp_model_kwargs": {},
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
}
|