phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-25 02:34:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add streaming_caching support

Browse source
This commit is contained in:
ishaan-jaff 2023-08-28 19:17:53 -07:00
parent 8af6d967eb
commit fbef73d043
3 changed files with 48 additions and 44 deletions
Download patch file Download diff file Expand all files Collapse all files

19
litellm/caching.py
View file

@ -1,4 +1,5 @@
import litellm
import time
def get_prompt(*args, **kwargs):
# make this safe checks, it should not throw any exceptions
if len(args) > 1:
@ -30,9 +31,9 @@ class InMemoryCache():
self.cache_dict = {}
def set_cache(self, key, value):
print("in set cache for inmem")
#print("in set cache for inmem")
self.cache_dict[key] = value
print(self.cache_dict)
#print(self.cache_dict)
def get_cache(self, key):
#print("in get cache for inmem")
@ -65,11 +66,23 @@ class Cache():
return None
return cache_key
def generate_streaming_content(self, content):
chunk_size = 5 # Adjust the chunk size as needed
for i in range(0, len(content), chunk_size):
yield {'choices': [{'delta': {'role': 'assistant', 'content': content[i:i+chunk_size]}}]}
time.sleep(0.02)
def get_cache(self, *args, **kwargs):
try: # never block execution
cache_key = self.get_cache_key(*args, **kwargs)
if cache_key is not None:
return self.cache.get_cache(cache_key)
cached_result = self.cache.get_cache(cache_key)
if cached_result != None and 'stream' in kwargs and kwargs['stream'] == True:
# if streaming is true and we got a cache hit, return a generator
#print("cache hit and stream=True")
#print(cached_result)
return self.generate_streaming_content(cached_result["choices"][0]['message']['content'])
return cached_result
except:
return None

53
litellm/tests/test_caching.py
View file

@ -152,35 +152,32 @@ def test_embedding_caching():
# test_embedding_caching()
# # test caching with streaming
# messages = [{"role": "user", "content": "hello gm who are u"}]
# def test_caching_v2_stream():
# try:
# litellm.cache = Cache()
# # litellm.token="ishaan@berri.ai"
# response1 = completion(model="gpt-3.5-turbo", messages=messages, stream=True)
# for chunk in response1:
# #
# pass
# # print("chunk")
# pass
# # response1_id = chunk['id']
# test caching with streaming
messages = [{"role": "user", "content": "tell me a story in 2 sentences"}]
def test_caching_v2_stream():
try:
litellm.cache = Cache()
# litellm.token="ishaan@berri.ai"
response1 = completion(model="gpt-3.5-turbo", messages=messages, stream=True)
result_string = ""
for chunk in response1:
print(chunk)
result_string+=chunk['choices'][0]['delta']['content']
# response1_id = chunk['id']
# # response2 = completion(model="gpt-3.5-turbo", messages=messages, stream=True)
# # for chunk in response2:
# # #print(chunk)
# # response2_id = chunk['id']
# # print(f"response1: {response1}")
# # print(f"response2: {response2}")
# # litellm.cache = None # disable cache
# # if response2_id != response1_id:
# # print(f"response1: {response1_id}")
# # print(f"response2: {response2_id}")
# # pytest.fail(f"Error occurred: {e}")
# except Exception as e:
# print(f"error occurred: {traceback.format_exc()}")
# pytest.fail(f"Error occurred: {e}")
result2_string=""
response2 = completion(model="gpt-3.5-turbo", messages=messages, stream=True)
for chunk in response2:
print(chunk)
result2_string+=chunk['choices'][0]['delta']['content']
if result_string != result2_string:
print(result_string)
print(result2_string)
pytest.fail(f"Error occurred: Caching with streaming failed, strings diff")
except Exception as e:
print(f"error occurred: {traceback.format_exc()}")
pytest.fail(f"Error occurred: {e}")
# test_caching_v2_stream()

20
litellm/utils.py
View file

@ -70,7 +70,7 @@ last_fetched_at_keys = None
class Message(OpenAIObject):
def __init__(self, content=" ", role="assistant", **params):
def __init__(self, content="default", role="assistant", **params):
super(Message, self).__init__(**params)
self.content = content
self.role = role
@ -287,24 +287,18 @@ class Logging:
)
if callback == "cache":
try:
#print("in cache callback2", self.stream)
#print(original_response)
#print(self.model_call_details)
if litellm.cache != None:
if litellm.cache != None and self.model_call_details.get('optional_params', {}).get('stream', False) == True:
if self.litellm_params["stream_response"] == None:
self.litellm_params["stream_response"] = ModelResponse()
else:
#self.litellm_call_id["stream_response"]["id"] = self.litellm_params["litellm_call_id"]
self.litellm_params["stream_response"]["choices"][0]["message"]["content"] += original_response
#print("cache is not none")
# convert original_response to format of Model Object
# Set the model
if self.litellm_params["stream_response"]["choices"][0]["message"]["content"] == "default":
self.litellm_params["stream_response"]["choices"][0]["message"]["content"] = original_response # handle first try
else:
self.litellm_params["stream_response"]["choices"][0]["message"]["content"] += original_response
litellm.cache.add_cache(self.litellm_params["stream_response"], **self.model_call_details)
#print(self.litellm_params["stream_response"])
except Exception as e:
print("got exception")
print(e)
pass
except:
print_verbose(
f"LiteLLM.LoggingError: [Non-Blocking] Exception occurred while post-call logging with integrations {traceback.format_exc()}"