diff --git a/examples/training/train_ner.py b/examples/training/train_ner.py
index 220244b93..bcc087d07 100644
--- a/examples/training/train_ner.py
+++ b/examples/training/train_ner.py
@@ -8,6 +8,12 @@ from spacy.pipeline import EntityRecognizer
 from spacy.gold import GoldParse
 from spacy.tagger import Tagger
 
+ 
+try:
+    unicode
+except:
+    unicode = str
+
 
 def train_ner(nlp, train_data, entity_types):
     # Add new words to vocab.
@@ -24,7 +30,6 @@ def train_ner(nlp, train_data, entity_types):
             doc = nlp.make_doc(raw_text)
             gold = GoldParse(doc, entities=entity_offsets)
             ner.update(doc, gold)
-    ner.model.end_training()
     return ner
 
 def save_model(ner, model_dir):
@@ -33,8 +38,11 @@ def save_model(ner, model_dir):
         model_dir.mkdir()
     assert model_dir.is_dir()
 
-    with (model_dir / 'config.json').open('w') as file_:
-        json.dump(ner.cfg, file_)
+    with (model_dir / 'config.json').open('wb') as file_:
+        data = json.dumps(ner.cfg)
+        if isinstance(data, unicode):
+            data = data.encode('utf8')
+        file_.write(data)
     ner.model.dump(str(model_dir / 'model'))
     if not (model_dir / 'vocab').exists():
         (model_dir / 'vocab').mkdir()