Tokenizer fix (#1)

- Tokenizer fix (2bc192c0384d35cb1821ba2f14f8d33f7135f02f)


Co-authored-by: Boris Malashenko <[email protected]>

tokenizer.json

@@ -85,8 +85,8 @@
   "pre_tokenizer": {
     "type": "Metaspace",
     "replacement": "▁",
-    "prepend_scheme": "always",
-    "split": true
+    "add_prefix_space": true,
+    "prepend_scheme": "always"
   },
   "post_processor": {
     "type": "TemplateProcessing",
@@ -172,8 +172,8 @@
   "decoder": {
     "type": "Metaspace",
     "replacement": "▁",
-    "prepend_scheme": "always",
-    "split": true
+    "add_prefix_space": true,
+    "prepend_scheme": "always"
   },
   "model": {
     "type": "Unigram",
@@ -184846,4 +184846,4 @@
     ],
     "byte_fallback": false
   }
-}
+}