Upload processor
Browse files- tokenizer_config.json +4 -0
tokenizer_config.json
CHANGED
@@ -1204,8 +1204,11 @@
|
|
1204 |
],
|
1205 |
"clean_up_tokenization_spaces": true,
|
1206 |
"eos_token": "</s>",
|
|
|
|
|
1207 |
"model_max_length": 1000000000000000019884624838656,
|
1208 |
"only_label_first_subword": true,
|
|
|
1209 |
"pad_token": "<pad>",
|
1210 |
"pad_token_box": [
|
1211 |
0,
|
@@ -1222,6 +1225,7 @@
|
|
1222 |
1000,
|
1223 |
1000
|
1224 |
],
|
|
|
1225 |
"tokenizer_class": "UdopTokenizer",
|
1226 |
"unk_token": "<unk>"
|
1227 |
}
|
|
|
1204 |
],
|
1205 |
"clean_up_tokenization_spaces": true,
|
1206 |
"eos_token": "</s>",
|
1207 |
+
"extra_ids": 100,
|
1208 |
+
"loc_extra_ids": 501,
|
1209 |
"model_max_length": 1000000000000000019884624838656,
|
1210 |
"only_label_first_subword": true,
|
1211 |
+
"other_extra_ids": 200,
|
1212 |
"pad_token": "<pad>",
|
1213 |
"pad_token_box": [
|
1214 |
0,
|
|
|
1225 |
1000,
|
1226 |
1000
|
1227 |
],
|
1228 |
+
"sp_model_kwargs": {},
|
1229 |
"tokenizer_class": "UdopTokenizer",
|
1230 |
"unk_token": "<unk>"
|
1231 |
}
|