fix tokenization config
Browse files- tokenizer_config.json +28 -6
tokenizer_config.json
CHANGED
|
@@ -1,8 +1,30 @@
|
|
| 1 |
{
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
}
|
|
|
|
|
|
| 1 |
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"eos_token": "</s>",
|
| 4 |
+
"unk_token": "<unk>",
|
| 5 |
+
"sep_token": "</s>",
|
| 6 |
+
"cls_token": "<s>",
|
| 7 |
+
"pad_token": "<pad>",
|
| 8 |
+
"mask_token": {
|
| 9 |
+
"content": "<mask>",
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"lstrip": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"normalized": true,
|
| 14 |
+
"__type": "AddedToken"
|
| 15 |
+
},
|
| 16 |
+
"sp_model_kwargs": {},
|
| 17 |
+
"do_lower_case": false,
|
| 18 |
+
"do_word_tokenize": true,
|
| 19 |
+
"do_subword_tokenize": true,
|
| 20 |
+
"word_tokenizer_type": "mecab",
|
| 21 |
+
"subword_tokenizer_type": "bpe",
|
| 22 |
+
"never_split": null,
|
| 23 |
+
"mecab_kwargs": {
|
| 24 |
+
"mecab_dic": "unidic_lite"
|
| 25 |
+
},
|
| 26 |
+
"special_tokens_map_file": null,
|
| 27 |
+
"tokenizer_file": null,
|
| 28 |
+
"tokenizer_class": "RobertaJapaneseTokenizer"
|
| 29 |
}
|
| 30 |
+
|