KoichiYasuoka
commited on
Commit
•
db3a38a
1
Parent(s):
0fb9b92
Whitespace
Browse files- tokenizer.json +12 -5
tokenizer.json
CHANGED
@@ -33,10 +33,17 @@
|
|
33 |
],
|
34 |
"normalizer": null,
|
35 |
"pre_tokenizer": {
|
36 |
-
"type": "
|
37 |
-
"
|
38 |
-
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
},
|
41 |
"post_processor": {
|
42 |
"type": "ByteLevel",
|
@@ -11828,4 +11835,4 @@
|
|
11828 |
"ìĭ ł"
|
11829 |
]
|
11830 |
}
|
11831 |
-
}
|
|
|
33 |
],
|
34 |
"normalizer": null,
|
35 |
"pre_tokenizer": {
|
36 |
+
"type": "Sequence",
|
37 |
+
"pretokenizers": [
|
38 |
+
{
|
39 |
+
"type": "Whitespace"
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"type": "ByteLevel",
|
43 |
+
"add_prefix_space": false,
|
44 |
+
"trim_offsets": true,
|
45 |
+
"use_regex": true
|
46 |
+
}]
|
47 |
},
|
48 |
"post_processor": {
|
49 |
"type": "ByteLevel",
|
|
|
11835 |
"ìĭ ł"
|
11836 |
]
|
11837 |
}
|
11838 |
+
}
|