phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

fix: setting cache responses on proxy

Browse source
This commit is contained in:
Krrish Dholakia 2023-12-07 20:39:40 -08:00
parent f5afc429b3
commit 2f1c5aa0c7
2 changed files with 7 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
litellm/proxy/proxy_server.py
View file

@ -460,6 +460,9 @@ def load_router_config(router: Optional[litellm.Router], config_file_path: str):
print_verbose(f"Loaded config YAML (api_key and environment_variables are not shown):\n{json.dumps(printed_yaml, indent=2)}")
## ROUTER CONFIG
cache_responses = False
## ENVIRONMENT VARIABLES
environment_variables = config.get('environment_variables', None)
if environment_variables:
@ -514,6 +517,7 @@ def load_router_config(router: Optional[litellm.Router], config_file_path: str):
print(f"{blue_color_code}\nSetting Cache on Proxy")
from litellm.caching import Cache
cache_type = value["type"]
cache_responses = True
cache_host = litellm.get_secret("REDIS_HOST", None)
cache_port = litellm.get_secret("REDIS_PORT", None)
cache_password = litellm.get_secret("REDIS_PASSWORD", None)
@ -565,7 +569,7 @@ def load_router_config(router: Optional[litellm.Router], config_file_path: str):
## MODEL LIST
model_list = config.get('model_list', None)
if model_list:
router = litellm.Router(model_list=model_list, num_retries=3)
router = litellm.Router(model_list=model_list, num_retries=3, cache_responses=cache_responses)
print(f"\033[32mLiteLLM: Proxy initialized with Config, Set models:\033[0m")
for model in model_list:
print(f"\033[32m {model.get('model_name', '')}\033[0m")

4
litellm/router.py
View file

@ -53,7 +53,7 @@ class Router:
```
"""
model_names: List = []
cache_responses: Optional[bool] = None
cache_responses: Optional[bool] = False
default_cache_time_seconds: int = 1 * 60 * 60 # 1 hour
num_retries: int = 0
tenacity = None
@ -65,7 +65,7 @@ class Router:
redis_host: Optional[str] = None,
redis_port: Optional[int] = None,
redis_password: Optional[str] = None,
cache_responses: Optional[bool] = None,
cache_responses: Optional[bool] = False,
cache_kwargs: dict = {}, # additional kwargs to pass to RedisCache (see caching.py)
## RELIABILITY ##
num_retries: int = 0,