phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-27 03:34:10 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

(feat) add response_time to StandardLoggingPayload - logged on datadog, gcs_bucket, s3_bucket etc (#7199)

Browse source 
* feat - add response_time to slp

* test_get_response_time

* docs slp

* fix test_datadog_logging_http_request
This commit is contained in:
Ishaan Jaff 2024-12-12 12:04:43 -08:00 • committed by GitHub
parent 04138c2df7
commit 2185587b4d
6 changed files with 73 additions and 40 deletions
Download patch file Download diff file Expand all files Collapse all files

31
litellm/litellm_core_utils/litellm_logging.py
View file

@ -2729,6 +2729,30 @@ class StandardLoggingPayloadSetup:
return api_base.rstrip("/")
return api_base
@staticmethod
def get_response_time(
start_time_float: float,
end_time_float: float,
completion_start_time_float: float,
stream: bool,
) -> float:
"""
Get the response time for the LLM response
Args:
start_time_float: float - start time of the LLM call
end_time_float: float - end time of the LLM call
completion_start_time_float: float - time to first token of the LLM response (for streaming responses)
stream: bool - True when a stream response is returned
Returns:
float: The response time for the LLM response
"""
if stream is True:
return completion_start_time_float - start_time_float
else:
return end_time_float - start_time_float
def get_standard_logging_object_payload(
kwargs: Optional[dict],
@ -2802,6 +2826,12 @@ def get_standard_logging_object_payload(
completion_start_time=completion_start_time,
)
)
response_time = StandardLoggingPayloadSetup.get_response_time(
start_time_float=start_time_float,
end_time_float=end_time_float,
completion_start_time_float=completion_start_time_float,
stream=kwargs.get("stream", False),
)
# clean up litellm hidden params
clean_hidden_params = StandardLoggingPayloadSetup.get_hidden_params(
hidden_params
@ -2850,6 +2880,7 @@ def get_standard_logging_object_payload(
startTime=start_time_float,
endTime=end_time_float,
completionStartTime=completion_start_time_float,
response_time=response_time,
model=kwargs.get("model", "") or "",
metadata=clean_metadata,
cache_key=clean_hidden_params["cache_key"],