feat: add observability logging for hybrid search mode

nemo_retriever/src/nemo_retriever/examples/batch_pipeline.py

@@ -589,6 +589,9 @@ def main(
         lancedb_uri = str(Path(lancedb_uri).expanduser().resolve())
         _ensure_lancedb_table(lancedb_uri, LANCEDB_TABLE)
 
+        if hybrid:
+            print("[hybrid] Hybrid search enabled (dense + FTS).")
+
         # Remote endpoints don't need local model GPUs for their stage.
         if page_elements_invoke_url and float(gpu_page_elements) != 0.0:
             print(
@@ -870,6 +873,7 @@ def main(
             hybrid=hybrid,
         )
 
+        print(f"[recall] Search mode: {'hybrid (dense + FTS + RRF rerank)' if hybrid else 'vector-only'}")
         _df_query, _gold, _raw_hits, _retrieved_keys, metrics = retrieve_and_score(query_csv=query_csv, cfg=cfg)
 
         if not no_recall_details:

nemo_retriever/src/nemo_retriever/ingest_modes/batch.py

@@ -1101,6 +1101,7 @@ def _create_lancedb_index(self) -> None:
             fts_language = str(kw.get("fts_language", "English"))
             try:
                 table.create_fts_index(text_column, language=fts_language)
+                print(f"[hybrid] FTS index created on column {text_column!r} (language={fts_language!r})")
             except Exception as e:
                 print(
                     f"Warning: FTS index creation failed on column {text_column!r} (continuing with vector-only): {e}"
@@ -1109,4 +1110,5 @@ def _create_lancedb_index(self) -> None:
         for index_stub in table.list_indices():
             table.wait_for_index([index_stub.name], timeout=timedelta(seconds=600))
 
-        print(f"Wrote {n_vecs} rows to LanceDB uri={lancedb_uri!r} table={table_name!r}")
+        index_names = [idx.name for idx in table.list_indices()]
+        print(f"Wrote {n_vecs} rows to LanceDB uri={lancedb_uri!r} table={table_name!r} indices={index_names}")

nemo_retriever/src/nemo_retriever/recall/core.py

@@ -196,6 +196,8 @@ def _search_lancedb(
         if effective_nprobes <= 0:
             effective_nprobes = 16  # safe fallback matching default index config
 
+    if hybrid:
+        logger.info("Recall search using hybrid mode (dense + FTS + RRF rerank)")
     results: List[List[Dict[str, Any]]] = []
     for i, v in enumerate(query_vectors):
         q = np.asarray(v, dtype="float32")