test

backend/danswer/document_index/vespa/index.py

@@ -451,6 +451,7 @@ def _query_vespa(query_params: Mapping[str, str | int]) -> list[InferenceChunk]:
 
     response_json = response.json()
     logger.debug("Response: %s", response_json)
+    logger.info("timing info: %s", response_json.get("timing"))
     hits = response_json["root"].get("children", [])
 
     for hit in hits:

backend/scripts/benchmark_search_isolated.py

@@ -141,7 +141,7 @@ def _measure_vespa_latency(filters: dict = {}):
     )
     query = generate_random_sentence()
     query_embedding = embed_query(query)
-    num_to_retrieve = 50
+    num_to_retrieve = 1
     params: dict[str, str | int] = {
         "yql": yql,
         "query": query,