Commit
·
f01df5d
1
Parent(s):
2225673
tokenizer improved
Browse files- tokenizer.json +3 -1
tokenizer.json
CHANGED
|
@@ -50,6 +50,7 @@
|
|
| 50 |
"pre_tokenizer": {
|
| 51 |
"type": "Metaspace",
|
| 52 |
"replacement": "▁",
|
|
|
|
| 53 |
"prepend_scheme": "always",
|
| 54 |
"split": true
|
| 55 |
},
|
|
@@ -57,6 +58,7 @@
|
|
| 57 |
"decoder": {
|
| 58 |
"type": "Metaspace",
|
| 59 |
"replacement": "▁",
|
|
|
|
| 60 |
"prepend_scheme": "always",
|
| 61 |
"split": true
|
| 62 |
},
|
|
@@ -128067,4 +128069,4 @@
|
|
| 128067 |
],
|
| 128068 |
"byte_fallback": false
|
| 128069 |
}
|
| 128070 |
-
}
|
|
|
|
| 50 |
"pre_tokenizer": {
|
| 51 |
"type": "Metaspace",
|
| 52 |
"replacement": "▁",
|
| 53 |
+
"add_prefix_space": true,
|
| 54 |
"prepend_scheme": "always",
|
| 55 |
"split": true
|
| 56 |
},
|
|
|
|
| 58 |
"decoder": {
|
| 59 |
"type": "Metaspace",
|
| 60 |
"replacement": "▁",
|
| 61 |
+
"add_prefix_space": true,
|
| 62 |
"prepend_scheme": "always",
|
| 63 |
"split": true
|
| 64 |
},
|
|
|
|
| 128069 |
],
|
| 128070 |
"byte_fallback": false
|
| 128071 |
}
|
| 128072 |
+
}
|