refactor: generated_data as list

Remove redundant conversions from None to empty list. Signed-off-by: Costa Shulyupin <[email protected]>
instructlab · Nov 18, 2024 · 03bdf9e · 03bdf9e
1 parent d382166
commit 03bdf9e
Showing 1 changed file with 3 additions and 9 deletions.
diff --git a/src/instructlab/sdg/generate_data.py b/src/instructlab/sdg/generate_data.py
@@ -409,7 +409,7 @@ def generate_data(
             "Synthesizing new instructions. If you aren't satisfied with the generated instructions, interrupt training (Ctrl-C) and try adjusting your YAML files. Adding more examples may help."
         )
 
-    generated_data = None
+    generated_data = []
     empty_sdg_leaf_nodes = []
     for leaf_node in leaf_nodes.values():
         is_knowledge = False
@@ -444,11 +444,8 @@ def generate_data(
             empty_sdg_leaf_nodes.append(leaf_node_path)
             logger.warning("Empty dataset for qna node: %s", leaf_node_path)
             continue
-        generated_data = (
-            [new_generated_data]
-            if generated_data is None
-            else generated_data + [new_generated_data]
-        )
+            generated_data.append(new_generated_data)
+
         logger.info("Generated %d samples", len(generated_data))
         logger.debug("Generated data: %s", generated_data)
 
@@ -469,9 +466,6 @@ def generate_data(
             use_legacy_pretraining_format,
         )
 
-    if generated_data is None:
-        generated_data = []
-
     _gen_train_data(
         generated_data,
         os.path.join(output_dir, output_file_train),