diff --git a/research/skip_thoughts/skip_thoughts/data/preprocess_dataset.py b/research/skip_thoughts/skip_thoughts/data/preprocess_dataset.py
index dfca7e8468971bcefe709cf0338697397790c5fd..b6f304f5301808177ae0f8ee59fb466d285b9217 100644
--- a/research/skip_thoughts/skip_thoughts/data/preprocess_dataset.py
+++ b/research/skip_thoughts/skip_thoughts/data/preprocess_dataset.py
@@ -121,8 +121,8 @@ def _build_vocabulary(input_files):
 
   tf.logging.info("Processed %d sentences total", num)
 
-  words = wordcount.keys()
-  freqs = wordcount.values()
+  words = list(wordcount)
+  freqs = list(wordcount.values())
   sorted_indices = np.argsort(freqs)[::-1]
 
   vocab = collections.OrderedDict()