diff --git a/file_utils/split_read_mini_batch.py b/file_utils/split_read_mini_batch.py
new file mode 100644
index 0000000..18330ec
--- /dev/null
+++ b/file_utils/split_read_mini_batch.py
@@ -0,0 +1,29 @@
+import os
+import sys
+
+def split_file_data(file_path, num_lines_per_file):
+    f = open(file_path)
+    rows = []
+    count = 0
+    batch_size = 50000
+    batch_id = 0
+    for row in f:
+        rows.append(row)
+        count += 1
+        if count == batch_size:
+            n_path, n_ext = os.path.splitext(file_path)
+            w = open(n_path + '-' + str(batch_id).rjust(5, '0') + n_ext, 'w')
+            for r in rows:
+                w.write(r)
+            w.close()
+            rows = []
+            count = 0
+            batch_id += 1
+
+
+if __name__ == '__main__':
+  data_dir_name = sys.argv[1]
+  files = ["edge.csv", "node-feat.csv", "node-label.csv"]
+  for file_name in files:
+     file_path = os.path.join(data_dir_name, file_name)
+     split_file_data(file_path, 50000)