Upload src/codewraith/app/retriever.py with huggingface_hub

src/codewraith/app/retriever.py

@@ -68,7 +68,7 @@ class SpecRetriever:
 
     def build_index(
         self,
-        dataset_path: str = "data/training_pairs_clean.jsonl",
+        dataset_path: str = "data/training_pairs_v7_clean.jsonl",
         batch_size: int = 50,
     ) -> int:
         """Build the vector index from training pairs.
@@ -196,7 +196,7 @@ class SpecRetriever:
         return "".join(parts)
 
 
-def build_index(dataset_path: str = "data/training_pairs_clean.jsonl") -> None:
+def build_index(dataset_path: str = "data/training_pairs_v7_clean.jsonl") -> None:
     """CLI entry point for building the vector index."""
     retriever = SpecRetriever()
     retriever.build_index(dataset_path)