diff --git a/buffalo/data/fileio.hpp b/buffalo/data/fileio.hpp
index ad57cff..49b2da7 100644
--- a/buffalo/data/fileio.hpp
+++ b/buffalo/data/fileio.hpp
@@ -322,7 +322,7 @@ vector<string> _sort_and_compressed_binarization(
         records.insert(end(records), begin(v), end_it);
     }
 
-    assert(records.size == total_lines);
+    assert(records.size() == total_lines);
 
     omp_set_num_threads(num_workers);
 
diff --git a/buffalo/data/stream.py b/buffalo/data/stream.py
index 3cdd283..3dc1fdd 100644
--- a/buffalo/data/stream.py
+++ b/buffalo/data/stream.py
@@ -105,6 +105,8 @@ def get_max_column_length(fname):
         with open(main_path) as fin:
             for line in log.ProgressBar(level=log.DEBUG, iterable=fin):
                 data = line.strip().split()
+                if not data:
+                    continue
                 if not iid_path:
                     itemids |= set(data)
 
@@ -246,7 +248,7 @@ def _create_working_data(self, db, stream_main_path, itemids,
                         for col in train_data:
                             w.write(f"{user} {col} 1\n")
                         for col in vali_data:
-                            vali_lines.append(f"{user} {col} {val}")
+                            vali_lines.append(f"{user} {col} 1")
                     else:
                         for col, val in Counter(train_data).items():
                             w.write(f"{user} {col} {val}\n")