KhimNguyen commited on
Commit
d42b110
1 Parent(s): 18c55ef

Upload tokenizer

Browse files
Files changed (1) hide show
  1. tokenizer.json +4 -4
tokenizer.json CHANGED
@@ -964,8 +964,8 @@
964
  "pre_tokenizer": {
965
  "type": "Metaspace",
966
  "replacement": "▁",
967
- "add_prefix_space": true,
968
- "prepend_scheme": "always"
969
  },
970
  "post_processor": {
971
  "type": "TemplateProcessing",
@@ -1024,8 +1024,8 @@
1024
  "decoder": {
1025
  "type": "Metaspace",
1026
  "replacement": "▁",
1027
- "add_prefix_space": true,
1028
- "prepend_scheme": "always"
1029
  },
1030
  "model": {
1031
  "type": "Unigram",
 
964
  "pre_tokenizer": {
965
  "type": "Metaspace",
966
  "replacement": "▁",
967
+ "prepend_scheme": "always",
968
+ "split": true
969
  },
970
  "post_processor": {
971
  "type": "TemplateProcessing",
 
1024
  "decoder": {
1025
  "type": "Metaspace",
1026
  "replacement": "▁",
1027
+ "prepend_scheme": "always",
1028
+ "split": true
1029
  },
1030
  "model": {
1031
  "type": "Unigram",