gabri3l commited on
Commit
f0debae
1 Parent(s): baf0c95

Upload tokenizer

Browse files
added_tokens.json CHANGED
@@ -1,5 +1,4 @@
1
  {
2
- "<nu>": 128104,
3
  "__af__": 128004,
4
  "__am__": 128005,
5
  "__ar__": 128006,
@@ -66,6 +65,7 @@
66
  "__nl__": 128067,
67
  "__no__": 128068,
68
  "__ns__": 128069,
 
69
  "__oc__": 128070,
70
  "__or__": 128071,
71
  "__pa__": 128072,
 
1
  {
 
2
  "__af__": 128004,
3
  "__am__": 128005,
4
  "__ar__": 128006,
 
65
  "__nl__": 128067,
66
  "__no__": 128068,
67
  "__ns__": 128069,
68
+ "__nu__": 128104,
69
  "__oc__": 128070,
70
  "__or__": 128071,
71
  "__pa__": 128072,
special_tokens_map.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "additional_special_tokens": [
3
  {
4
- "content": "<nu>",
5
  "lstrip": false,
6
  "normalized": false,
7
  "rstrip": false,
 
1
  {
2
  "additional_special_tokens": [
3
  {
4
+ "content": "__nu__",
5
  "lstrip": false,
6
  "normalized": false,
7
  "rstrip": false,
tokenizer_config.json CHANGED
@@ -833,7 +833,7 @@
833
  "special": true
834
  },
835
  "128104": {
836
- "content": "<nu>",
837
  "lstrip": false,
838
  "normalized": false,
839
  "rstrip": false,
@@ -842,7 +842,7 @@
842
  }
843
  },
844
  "additional_special_tokens": [
845
- "<nu>"
846
  ],
847
  "bos_token": "<s>",
848
  "clean_up_tokenization_spaces": true,
 
833
  "special": true
834
  },
835
  "128104": {
836
+ "content": "__nu__",
837
  "lstrip": false,
838
  "normalized": false,
839
  "rstrip": false,
 
842
  }
843
  },
844
  "additional_special_tokens": [
845
+ "__nu__"
846
  ],
847
  "bos_token": "<s>",
848
  "clean_up_tokenization_spaces": true,