From d5f3a88e74adbf3c605e82258d7d611cd70e0c4c Mon Sep 17 00:00:00 2001 From: norrishuang <12380647@qq.com> Date: Wed, 15 Oct 2025 06:44:27 +0000 Subject: [PATCH 1/2] awsopensearch add normalize for cosine similarity, solved isuue of recall rate with Cohere dataset --- .../backend/clients/aws_opensearch/aws_opensearch.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py b/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py index 97b3726ed..f571c83d1 100644 --- a/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py +++ b/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py @@ -64,6 +64,12 @@ def __init__( self._update_ef_search_before_search(client) self._load_graphs_to_memory(client) + def need_normalize_cosine(self) -> bool: + if self.case_config.metric_type.upper() == 'COSINE': + log.info("cosine dataset need normalize.") + return True + return False + def _create_index(self, client: OpenSearch) -> None: ef_search_value = self.case_config.ef_search log.info(f"Creating index with ef_search: {ef_search_value}") From 7328bdb7470060b6b76973211ca421ab68796d67 Mon Sep 17 00:00:00 2001 From: norrishuang <12380647@qq.com> Date: Wed, 15 Oct 2025 13:05:06 +0000 Subject: [PATCH 2/2] fix format issue --- .../backend/clients/aws_opensearch/aws_opensearch.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py b/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py index f571c83d1..2161c7c02 100644 --- a/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py +++ b/vectordb_bench/backend/clients/aws_opensearch/aws_opensearch.py @@ -65,11 +65,11 @@ def __init__( self._load_graphs_to_memory(client) def need_normalize_cosine(self) -> bool: - if self.case_config.metric_type.upper() == 'COSINE': + if self.case_config.metric_type.upper() == "COSINE": log.info("cosine dataset need normalize.") return True return False - + def _create_index(self, client: OpenSearch) -> None: ef_search_value = self.case_config.ef_search log.info(f"Creating index with ef_search: {ef_search_value}")