fix(utils.py): await async function in client wrapper

litellm/tests/test_async_fn.py

@@ -42,6 +42,7 @@ def test_get_response_streaming():
         user_message = "write a short poem in one sentence"
         messages = [{"content": user_message, "role": "user"}]
         try:
+            litellm.set_verbose = True
             response = await acompletion(model="command-nightly", messages=messages, stream=True)
             print(type(response))
 
@@ -65,7 +66,7 @@ def test_get_response_streaming():
     asyncio.run(test_async_call())
 
 
-# test_get_response_streaming()
+test_get_response_streaming()
 
 def test_get_response_non_openai_streaming():
     import asyncio

litellm/utils.py

@@ -1238,7 +1238,7 @@ def client(original_function):
                         else:
                             return cached_result
             # MODEL CALL
-            result = original_function(*args, **kwargs)
+            result = await original_function(*args, **kwargs)
             end_time = datetime.datetime.now()
             if "stream" in kwargs and kwargs["stream"] == True:
                 if "complete_response" in kwargs and kwargs["complete_response"] == True: 
@@ -1248,7 +1248,6 @@ def client(original_function):
                     return litellm.stream_chunk_builder(chunks)
                 else: 
                     return result
-            result = await result 
             # [OPTIONAL] ADD TO CACHE
             if litellm.caching or litellm.caching_with_models or litellm.cache != None: # user init a cache object
                 litellm.cache.add_cache(result, *args, **kwargs)
@@ -4459,7 +4458,6 @@ class CustomStreamWrapper:
             traceback.print_exc()
             raise e
 
-
     def handle_openai_text_completion_chunk(self, chunk):
         try: 
             str_line = chunk