fix(caching.py): fix print statements

litellm/caching.py

@@ -142,7 +142,9 @@ class RedisCache(BaseCache):
                 )
             except Exception as e:
                 # NON blocking - notify users Redis is throwing an exception
-                print_verbose("LiteLLM Caching: set() - Got exception from REDIS : ", e)
+                print_verbose(
+                    f"LiteLLM Redis Caching: async set() - Got exception from REDIS : {str(e)}"
+                )
 
     async def async_set_cache_pipeline(self, cache_list, ttl=None):
         """
@@ -905,8 +907,11 @@ class Cache:
 
         # for streaming, we use preset_cache_key. It's created in wrapper(), we do this because optional params like max_tokens, get transformed for bedrock -> max_new_tokens
         if kwargs.get("litellm_params", {}).get("preset_cache_key", None) is not None:
-            print_verbose(f"\nReturning preset cache key: {cache_key}")
+            _preset_cache_key = kwargs.get("litellm_params", {}).get(
-            return kwargs.get("litellm_params", {}).get("preset_cache_key", None)
+                "preset_cache_key", None
+            )
+            print_verbose(f"\nReturning preset cache key: {_preset_cache_key}")
+            return _preset_cache_key
 
         # sort kwargs by keys, since model: [gpt-4, temperature: 0.2, max_tokens: 200] == [temperature: 0.2, max_tokens: 200, model: gpt-4]
         completion_kwargs = [

litellm/proxy/_new_secret_config.yaml

@@ -4,6 +4,10 @@ model_list:
     model: openai/my-fake-model
     api_key: my-fake-key
     api_base: http://0.0.0.0:8080
+- model_name: gpt-3.5-turbo
+  litellm_params:
+    model: gpt-3.5-turbo-1106
+    api_key: os.environ/OPENAI_API_KEY
 
 general_settings:
   master_key: sk-1234

litellm/proxy/proxy_server.py

@@ -1112,15 +1112,25 @@ async def update_database(
                                 max_budget=max_user_budget,
                                 user_email=None,
                             )
+
                         else:
                             existing_user_obj.spend = (
                                 existing_user_obj.spend + response_cost
                             )
 
+                        user_object_json = {**existing_user_obj.json(exclude_none=True)}
+
+                        user_object_json["model_max_budget"] = json.dumps(
+                            user_object_json["model_max_budget"]
+                        )
+                        user_object_json["model_spend"] = json.dumps(
+                            user_object_json["model_spend"]
+                        )
+
                         await prisma_client.db.litellm_usertable.upsert(
                             where={"user_id": end_user_id},
                             data={
-                                "create": {**existing_user_obj.json(exclude_none=True)},
+                                "create": user_object_json,
                                 "update": {"spend": {"increment": response_cost}},
                             },
                         )