Split load and transform steps in examples for readability

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: J03D03

examples/04_enrich_with_commit_data.py

@@ -9,5 +9,6 @@
 
 if __name__ == "__main__":
     # Devign contains only 2 projects (FFmpeg, QEMU).
-    entries = transformations.add_commit_information_local(vfc_datasets.DevignDataset())
+    entries = vfc_datasets.DevignDataset()
+    entries = transformations.add_commit_information_local(entries)
     log_dataset_stats(entries)

examples/06_create_splits.py

@@ -9,8 +9,10 @@
 setup_logging("create_splits")
 
 if __name__ == "__main__":
+    entries = vfc_datasets.DevignDataset()
+
     create_random_split(
-        vfc_datasets.DevignDataset(),
+        entries,
         name="devign",
         output_path=Path(".data/splits"),
         seed=42,

examples/07_filter_by_language.py

@@ -10,7 +10,8 @@
 logger = logging.getLogger(__name__)
 
 if __name__ == "__main__":
-    entries = transformations.add_commit_information_local(vfc_datasets.DevignDataset())
+    entries = vfc_datasets.DevignDataset()
+    entries = transformations.add_commit_information_local(entries)
 
     logger.info("Entries before filtering: %d", len(entries))
     entries = transformations.filter_by_extension(entries, {"c", "h", "cpp", "hpp", "cc"})