From 80fc8050eb3e55cc6d0ee526e4fe46c5b0525b6f Mon Sep 17 00:00:00 2001 From: ishaan-jaff Date: Thu, 14 Dec 2023 13:24:10 +0530 Subject: [PATCH] (feat) proxy - model group alias --- litellm/proxy/proxy_server.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py index 8485765afb..3b77d4d010 100644 --- a/litellm/proxy/proxy_server.py +++ b/litellm/proxy/proxy_server.py @@ -1002,7 +1002,7 @@ async def chat_completion(request: Request, model: Optional[str] = None, user_ap response = await llm_router.acompletion(**data) elif llm_router is not None and data["model"] in llm_router.deployment_names: # model in router deployments, calling a specific deployment on the router response = await llm_router.acompletion(**data, specific_deployment = True) - elif llm_router is not None and litellm.model_group_alias_map is not None and data["model"] in litellm.model_group_alias_map: # model set in model_group_alias_map + elif llm_router is not None and llm_router.model_group_alias is not None and data["model"] in llm_router.model_group_alias: # model set in model_group_alias response = await llm_router.acompletion(**data) else: # router is not set response = await litellm.acompletion(**data) @@ -1100,6 +1100,8 @@ async def embeddings(request: Request, user_api_key_dict: UserAPIKeyAuth = Depen response = await llm_router.aembedding(**data) elif llm_router is not None and data["model"] in llm_router.deployment_names: # model in router deployments, calling a specific deployment on the router response = await llm_router.aembedding(**data, specific_deployment = True) + elif llm_router is not None and llm_router.model_group_alias is not None and data["model"] in llm_router.model_group_alias: # model set in model_group_alias + response = await llm_router.aembedding(**data) # ensure this goes the llm_router, router will do the correct alias mapping else: response = await litellm.aembedding(**data) background_tasks.add_task(log_input_output, request, response) # background task for logging to OTEL