Fix pickle error: Dense index uses its own caching

app_retrieval_cached.py

@@ -80,7 +80,7 @@ def _get_cache_key(corpora_config: Dict[str, dict]) -> str:
 
 
 class CachedRetriever:
-    """Retriever with disk caching for BM25 and Dense indexes"""
+    """Retriever with disk caching for BM25 and documents (Dense has its own caching)"""
     
     def __init__(self, corpora_config: Dict[str, dict], use_reranker: bool = False):
         self.corpora_config = corpora_config
@@ -89,8 +89,8 @@ class CachedRetriever:
         
         # Cache file paths
         self.bm25_cache = CACHE_DIR / f"bm25_{self.cache_key}.pkl"
-        self.dense_cache = CACHE_DIR / f"dense_{self.cache_key}.pkl"
         self.docs_cache = CACHE_DIR / f"docs_{self.cache_key}.pkl"
+        # Note: Dense index uses its own caching in .cache/embeddings/
         
         # Load or build indexes
         self.docs_all = self._load_or_build_docs()
@@ -148,26 +148,12 @@ class CachedRetriever:
         return bm25_index
     
     def _load_or_build_dense(self) -> DenseIndex:
-        """Load Dense index from cache or build from scratch"""
-        if self.dense_cache.exists():
-            print(f"Loading Dense index from cache... ({self.dense_cache.name})")
-            try:
-                with open(self.dense_cache, 'rb') as f:
-                    dense_index = pickle.load(f)
-                print(f"  ✓ Dense index loaded from cache")
-                return dense_index
-            except Exception as e:
-                print(f"  ✗ Cache load failed: {e}")
-                print("  → Rebuilding Dense index...")
-        
-        print("Building Dense index from scratch (this takes 5-8 minutes)...")
+        """Build Dense index (note: Dense index has its own internal caching)"""
+        print("Initializing Dense index (uses internal caching)...")
+        # DenseIndex has its own caching system in .cache/embeddings/
+        # We don't need to pickle it - just let it build/load from its own cache
         dense_index = DenseIndex(self.docs_all)
-        
-        # Save to cache
-        print(f"Saving Dense index to cache...")
-        with open(self.dense_cache, 'wb') as f:
-            pickle.dump(dense_index, f)
-        
+        print(f"  ✓ Dense index ready")
         return dense_index