Merge pull request #4898 from RasaHQ/fix_file_order

Make order of files loaded consistent
RasaHQ · Dec 4, 2019 · acedba4 · acedba4
2 parents 93b0948 + 33ebe75
commit acedba4
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 1 deletion.
diff --git a/changelog/4898.bugfix.rst b/changelog/4898.bugfix.rst
@@ -0,0 +1 @@
+Training data files now get loaded in the same order (especially relevant to subdirectories) each time to ensure training consistency when using a random seed.
diff --git a/rasa/data.py b/rasa/data.py
@@ -103,7 +103,8 @@ def _find_core_nlu_files_in_directory(directory: Text,) -> Tuple[Set[Text], Set[
     nlu_data_files = set()
 
     for root, _, files in os.walk(directory):
-        for f in files:
+        # we sort the files here to ensure consistent order for repeatable training results
+        for f in sorted(files):
             full_path = os.path.join(root, f)
 
             if not _is_valid_filetype(full_path):