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

(fix) improve mem util

Browse source
This commit is contained in:
ishaan-jaff 2024-03-11 16:22:04 -07:00
parent 2f1899284c
commit 1bd3bb1128
2 changed files with 149 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

14
litellm/router.py
View file

@ -210,9 +210,6 @@ class Router:
self.context_window_fallbacks = (
context_window_fallbacks or litellm.context_window_fallbacks
)
self.model_exception_map: dict = (
{}
) # dict to store model: list exceptions. self.exceptions = {"gpt-3.5": ["API KEY Error", "Rate Limit Error", "good morning error"]}
self.total_calls: defaultdict = defaultdict(
int
) # dict to store total calls made to each model
@ -1487,17 +1484,6 @@ class Router:
self._set_cooldown_deployments(
deployment_id
) # setting deployment_id in cooldown deployments
if metadata:
deployment = metadata.get("deployment", None)
deployment_exceptions = self.model_exception_map.get(deployment, [])
deployment_exceptions.append(exception_str)
self.model_exception_map[deployment] = deployment_exceptions
verbose_router_logger.debug("\nEXCEPTION FOR DEPLOYMENTS\n")
verbose_router_logger.debug(self.model_exception_map)
for model in self.model_exception_map:
verbose_router_logger.debug(
f"Model {model} had {len(self.model_exception_map[model])} exception"
)
if custom_llm_provider:
model_name = f"{custom_llm_provider}/{model_name}"