fix(main.py): fix together ai text completion call

litellm/main.py

@@ -2950,7 +2950,9 @@ def embedding(
                     model=model,  # type: ignore
                     llm_provider="ollama",  # type: ignore
                 )
-            ollama_embeddings_fn = ollama.ollama_aembeddings if aembedding else ollama.ollama_embeddings
+            ollama_embeddings_fn = (
+                ollama.ollama_aembeddings if aembedding else ollama.ollama_embeddings
+            )
             response = ollama_embeddings_fn(
                 api_base=api_base,
                 model=model,
@@ -3094,6 +3096,7 @@ async def atext_completion(*args, **kwargs):
             or custom_llm_provider == "huggingface"
             or custom_llm_provider == "ollama"
             or custom_llm_provider == "vertex_ai"
+            or custom_llm_provider in litellm.openai_compatible_providers
         ):  # currently implemented aiohttp calls for just azure and openai, soon all.
             # Await normally
             response = await loop.run_in_executor(None, func_with_context)
@@ -3124,6 +3127,8 @@ async def atext_completion(*args, **kwargs):
             ## TRANSLATE CHAT TO TEXT FORMAT ##
             if isinstance(response, TextCompletionResponse):
                 return response
+            elif asyncio.iscoroutine(response):
+                response = await response
 
             text_completion_response = TextCompletionResponse()
             text_completion_response["id"] = response.get("id", None)

litellm/proxy/_super_secret_config.yaml

@@ -17,6 +17,9 @@ model_list:
 - model_name: gpt-4
   litellm_params:
     model: gpt-3.5-turbo
+- litellm_params:
+    model: together_ai/codellama/CodeLlama-13b-Instruct-hf
+  model_name: CodeLlama-13b-Instruct
 router_settings:
   num_retries: 0
   enable_pre_call_checks: true

litellm/tests/test_text_completion.py

@@ -3990,6 +3990,26 @@ def test_async_text_completion():
     asyncio.run(test_get_response())
 
 
+def test_async_text_completion_together_ai():
+    litellm.set_verbose = True
+    print("test_async_text_completion")
+
+    async def test_get_response():
+        try:
+            response = await litellm.atext_completion(
+                model="together_ai/codellama/CodeLlama-13b-Instruct-hf",
+                prompt="good morning",
+                max_tokens=10,
+            )
+            print(f"response: {response}")
+        except litellm.Timeout as e:
+            print(e)
+        except Exception as e:
+            pytest.fail("An unexpected error occurred")
+
+    asyncio.run(test_get_response())
+
+
 # test_async_text_completion()