diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py index c2d3d194ae..6f442f1ae3 100644 --- a/litellm/proxy/proxy_server.py +++ b/litellm/proxy/proxy_server.py @@ -1178,7 +1178,7 @@ class ProxyConfig: verbose_proxy_logger.debug(f"passed cache type={cache_type}") - if cache_type == "redis": + if cache_type == "redis" or cache_type == "redis-semantic": cache_host = litellm.get_secret("REDIS_HOST", None) cache_port = litellm.get_secret("REDIS_PORT", None) cache_password = litellm.get_secret("REDIS_PASSWORD", None) @@ -1205,6 +1205,9 @@ class ProxyConfig: f"{blue_color_code}Cache Password:{reset_color_code} {cache_password}" ) print() # noqa + if cache_type == "redis-semantic": + # by default this should always be async + cache_params.update({"redis_semantic_cache_use_async": True}) # users can pass os.environ/ variables on the proxy - we should read them from the env for key, value in cache_params.items(): diff --git a/requirements.txt b/requirements.txt index b0a49553d1..3ace5872ad 100644 --- a/requirements.txt +++ b/requirements.txt @@ -10,6 +10,7 @@ gunicorn==21.2.0 # server dep boto3==1.28.58 # aws bedrock/sagemaker calls redis==4.6.0 # caching redisvl==0.0.7 # semantic caching +numpy==1.24.3 # semantic caching prisma==0.11.0 # for db mangum==0.17.0 # for aws lambda functions google-generativeai==0.3.2 # for vertex ai calls