Commit
·
7eef22f
1
Parent(s):
af0ad52
Upload tokenizer
Browse files- added_tokens.json +0 -2
- tokenizer.json +0 -18
- tokenizer_config.json +0 -16
added_tokens.json
CHANGED
@@ -1,6 +1,4 @@
|
|
1 |
{
|
2 |
-
"<CS>": 50400,
|
3 |
-
"<EC>": 50401,
|
4 |
"<|extratoken_100|>": 50356,
|
5 |
"<|extratoken_101|>": 50357,
|
6 |
"<|extratoken_102|>": 50358,
|
|
|
1 |
{
|
|
|
|
|
2 |
"<|extratoken_100|>": 50356,
|
3 |
"<|extratoken_101|>": 50357,
|
4 |
"<|extratoken_102|>": 50358,
|
tokenizer.json
CHANGED
@@ -1298,24 +1298,6 @@
|
|
1298 |
"rstrip": false,
|
1299 |
"normalized": true,
|
1300 |
"special": false
|
1301 |
-
},
|
1302 |
-
{
|
1303 |
-
"id": 50400,
|
1304 |
-
"content": "<CS>",
|
1305 |
-
"single_word": false,
|
1306 |
-
"lstrip": false,
|
1307 |
-
"rstrip": false,
|
1308 |
-
"normalized": true,
|
1309 |
-
"special": false
|
1310 |
-
},
|
1311 |
-
{
|
1312 |
-
"id": 50401,
|
1313 |
-
"content": "<EC>",
|
1314 |
-
"single_word": false,
|
1315 |
-
"lstrip": false,
|
1316 |
-
"rstrip": false,
|
1317 |
-
"normalized": true,
|
1318 |
-
"special": false
|
1319 |
}
|
1320 |
],
|
1321 |
"normalizer": null,
|
|
|
1298 |
"rstrip": false,
|
1299 |
"normalized": true,
|
1300 |
"special": false
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1301 |
}
|
1302 |
],
|
1303 |
"normalizer": null,
|
tokenizer_config.json
CHANGED
@@ -1153,22 +1153,6 @@
|
|
1153 |
"rstrip": false,
|
1154 |
"single_word": false,
|
1155 |
"special": false
|
1156 |
-
},
|
1157 |
-
"50400": {
|
1158 |
-
"content": "<CS>",
|
1159 |
-
"lstrip": false,
|
1160 |
-
"normalized": true,
|
1161 |
-
"rstrip": false,
|
1162 |
-
"single_word": false,
|
1163 |
-
"special": false
|
1164 |
-
},
|
1165 |
-
"50401": {
|
1166 |
-
"content": "<EC>",
|
1167 |
-
"lstrip": false,
|
1168 |
-
"normalized": true,
|
1169 |
-
"rstrip": false,
|
1170 |
-
"single_word": false,
|
1171 |
-
"special": false
|
1172 |
}
|
1173 |
},
|
1174 |
"bos_token": "<|endoftext|>",
|
|
|
1153 |
"rstrip": false,
|
1154 |
"single_word": false,
|
1155 |
"special": false
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1156 |
}
|
1157 |
},
|
1158 |
"bos_token": "<|endoftext|>",
|