diff --git a/examples/training/conllu.py b/examples/training/conllu.py
index 50716a0e1..271a049c0 100644
--- a/examples/training/conllu.py
+++ b/examples/training/conllu.py
@@ -28,6 +28,8 @@ def prevent_bad_sentences(doc):
             token.is_sent_start = False
         elif not token.nbor(-1).is_punct:
             token.is_sent_start = False
+        elif token.nbor(-1).is_left_punct:
+            token.is_sent_start = False
     return doc
 
 
@@ -99,7 +101,7 @@ def read_data(nlp, conllu_file, text_file, raw_text=True, oracle_segments=False,
     # cs is conllu sent, ct is conllu token
     docs = []
     golds = []
-    for text, cd in zip(paragraphs, conllu):
+    for doc_id, (text, cd) in enumerate(zip(paragraphs, conllu)):
         doc_words = []
         doc_tags = []
         doc_heads = []
@@ -140,7 +142,7 @@ def read_data(nlp, conllu_file, text_file, raw_text=True, oracle_segments=False,
             golds.append(GoldParse(docs[-1], words=doc_words, tags=doc_tags,
                                    heads=doc_heads, deps=doc_deps,
                                    entities=doc_ents))
-        if limit and len(docs) >= limit:
+        if limit and doc_id >= limit:
             break
     return docs, golds
 
@@ -188,7 +190,14 @@ def parse_dev_data(nlp, text_loc, conllu_loc, oracle_segments=False,
         with open(conllu_loc) as conllu_file:
             docs, golds = read_data(nlp, conllu_file, text_file,
                                     oracle_segments=oracle_segments)
-    if not joint_sbd:
+    if joint_sbd:
+        sbd = nlp.create_pipe('sentencizer')
+        for doc in docs:
+            doc = sbd(doc)
+            for sent in doc.sents:
+                sent[0].is_sent_start = True
+        #docs = (prevent_bad_sentences(doc) for doc in docs)
+    else:
         sbd = nlp.create_pipe('sentencizer')
         for doc in docs:
             doc = sbd(doc)
@@ -245,7 +254,8 @@ def main(spacy_model, conllu_train_loc, text_train_loc, conllu_dev_loc, text_dev
     with open(conllu_train_loc) as conllu_file:
         with open(text_train_loc) as text_file:
             docs, golds = read_data(nlp, conllu_file, text_file,
-                                    oracle_segments=False, raw_text=True)
+                                    oracle_segments=True, raw_text=True,
+                                    limit=None)
     print("Create parser")
     nlp.add_pipe(nlp.create_pipe('parser'))
     nlp.add_pipe(nlp.create_pipe('tagger'))
@@ -266,7 +276,7 @@ def main(spacy_model, conllu_train_loc, text_train_loc, conllu_dev_loc, text_dev
     print("Begin training")
     # Batch size starts at 1 and grows, so that we make updates quickly
     # at the beginning of training.
-    batch_sizes = spacy.util.compounding(spacy.util.env_opt('batch_from', 1),
+    batch_sizes = spacy.util.compounding(spacy.util.env_opt('batch_from', 2),
                                    spacy.util.env_opt('batch_to', 2),
                                    spacy.util.env_opt('batch_compound', 1.001))
     for i in range(30):
@@ -278,6 +288,7 @@ def main(spacy_model, conllu_train_loc, text_train_loc, conllu_dev_loc, text_dev
                 if not batch:
                     continue
                 batch_docs, batch_gold = zip(*batch)
+                batch_docs = [prevent_bad_sentences(doc) for doc in batch_docs]
 
                 nlp.update(batch_docs, batch_gold, sgd=optimizer,
                            drop=0.2, losses=losses)
@@ -296,6 +307,5 @@ def main(spacy_model, conllu_train_loc, text_train_loc, conllu_dev_loc, text_dev
                 print_conllu(dev_docs, file_)
 
 
-
 if __name__ == '__main__':
     plac.call(main)