Commit
•
6fb664c
1
Parent(s):
d7a4711
Upload tokenizer
Browse files- tokenizer_config.json +4 -0
tokenizer_config.json
CHANGED
|
@@ -1867,12 +1867,16 @@
|
|
| 1867 |
"eos_token": "</s>",
|
| 1868 |
"legacy_behaviour": false,
|
| 1869 |
"mask_token": "<mask>",
|
|
|
|
| 1870 |
"model_max_length": 1024,
|
| 1871 |
"pad_token": "<pad>",
|
| 1872 |
"sep_token": "</s>",
|
| 1873 |
"sp_model_kwargs": {},
|
| 1874 |
"src_lang": "kor-Hang",
|
|
|
|
| 1875 |
"tgt_lang": "eng-Latn",
|
| 1876 |
"tokenizer_class": "NllbTokenizer",
|
|
|
|
|
|
|
| 1877 |
"unk_token": "<unk>"
|
| 1878 |
}
|
|
|
|
| 1867 |
"eos_token": "</s>",
|
| 1868 |
"legacy_behaviour": false,
|
| 1869 |
"mask_token": "<mask>",
|
| 1870 |
+
"max_length": 128,
|
| 1871 |
"model_max_length": 1024,
|
| 1872 |
"pad_token": "<pad>",
|
| 1873 |
"sep_token": "</s>",
|
| 1874 |
"sp_model_kwargs": {},
|
| 1875 |
"src_lang": "kor-Hang",
|
| 1876 |
+
"stride": 0,
|
| 1877 |
"tgt_lang": "eng-Latn",
|
| 1878 |
"tokenizer_class": "NllbTokenizer",
|
| 1879 |
+
"truncation_side": "right",
|
| 1880 |
+
"truncation_strategy": "longest_first",
|
| 1881 |
"unk_token": "<unk>"
|
| 1882 |
}
|