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

Merge pull request #3250 from BerriAI/litellm_caching_no_cache_fix

Browse source 
fix(utils.py): fix 'no-cache': true when caching is turned on
This commit is contained in:
Krish Dholakia 2024-04-23 19:57:07 -07:00 • committed by GitHub
commit 4acdde988f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 122 additions and 55 deletions
Download patch file Download diff file Expand all files Collapse all files

33
litellm/utils.py
View file

@ -2716,23 +2716,22 @@ def client(original_function):
# [OPTIONAL] CHECK CACHE
print_verbose(
f"kwargs[caching]: {kwargs.get('caching', False)}; litellm.cache: {litellm.cache}"
f"SYNC kwargs[caching]: {kwargs.get('caching', False)}; litellm.cache: {litellm.cache}; kwargs.get('cache')['no-cache']: {kwargs.get('cache', {}).get('no-cache', False)}"
)
# if caching is false or cache["no-cache"]==True, don't run this
if (
(
(
kwargs.get("caching", None) is None
and kwargs.get("cache", None) is None
and litellm.cache is not None
)
or kwargs.get("caching", False) == True
or (
kwargs.get("cache", None) is not None
and kwargs.get("cache", {}).get("no-cache", False) != True
(
kwargs.get("caching", None) is None
and litellm.cache is not None
)
or kwargs.get("caching", False) == True
)
and kwargs.get("cache", {}).get("no-cache", False) != True
)
and kwargs.get("aembedding", False) != True
and kwargs.get("atext_completion", False) != True
and kwargs.get("acompletion", False) != True
and kwargs.get("aimg_generation", False) != True
and kwargs.get("atranscription", False) != True
@ -3011,24 +3010,17 @@ def client(original_function):
)
# [OPTIONAL] CHECK CACHE
print_verbose(f"litellm.cache: {litellm.cache}")
print_verbose(
f"kwargs[caching]: {kwargs.get('caching', False)}; litellm.cache: {litellm.cache}"
f"ASYNC kwargs[caching]: {kwargs.get('caching', False)}; litellm.cache: {litellm.cache}; kwargs.get('cache'): {kwargs.get('cache', None)}"
)
# if caching is false, don't run this
final_embedding_cached_response = None
if (
(
kwargs.get("caching", None) is None
and kwargs.get("cache", None) is None
and litellm.cache is not None
)
(kwargs.get("caching", None) is None and litellm.cache is not None)
or kwargs.get("caching", False) == True
or (
kwargs.get("cache", None) is not None
and kwargs.get("cache").get("no-cache", False) != True
)
) and (
kwargs.get("cache", {}).get("no-cache", False) != True
): # allow users to control returning cached responses from the completion function
# checking cache
print_verbose("INSIDE CHECKING CACHE")
@ -3074,7 +3066,6 @@ def client(original_function):
preset_cache_key # for streaming calls, we need to pass the preset_cache_key
)
cached_result = litellm.cache.get_cache(*args, **kwargs)
if cached_result is not None and not isinstance(
cached_result, list
):