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

fix prometheus track input and output tokens (#5780)

Browse source
This commit is contained in:
Ishaan Jaff 2024-09-18 17:09:14 -07:00
parent 2a8eb492a1
commit 37be0088ff
1 changed files with 73 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

104
litellm/integrations/prometheus.py
View file

@ -120,6 +120,33 @@ class PrometheusLogger(CustomLogger):
], ],
) )
self.litellm_input_tokens_metric = Counter(
"litellm_input_tokens",
"Total number of input tokens from LLM requests",
labelnames=[
"end_user",
"hashed_api_key",
"api_key_alias",
"model",
"team",
"team_alias",
"user",
],
)
self.litellm_output_tokens_metric = Counter(
"litellm_output_tokens",
"Total number of output tokens from LLM requests",
labelnames=[
"end_user",
"hashed_api_key",
"api_key_alias",
"model",
"team",
"team_alias",
"user",
],
)
# Remaining Budget for Team # Remaining Budget for Team
self.litellm_remaining_team_budget_metric = Gauge( self.litellm_remaining_team_budget_metric = Gauge(
"litellm_remaining_team_budget_metric", "litellm_remaining_team_budget_metric",
@ -269,30 +296,31 @@ class PrometheusLogger(CustomLogger):
from litellm.proxy.common_utils.callback_utils import ( from litellm.proxy.common_utils.callback_utils import (
get_model_group_from_litellm_kwargs, get_model_group_from_litellm_kwargs,
) )
from litellm.types.utils import StandardLoggingPayload
verbose_logger.debug( verbose_logger.debug(
f"prometheus Logging - Enters success logging function for kwargs {kwargs}" f"prometheus Logging - Enters success logging function for kwargs {kwargs}"
) )
# unpack kwargs # unpack kwargs
standard_logging_payload: StandardLoggingPayload = kwargs.get(
"standard_logging_object", {}
)
model = kwargs.get("model", "") model = kwargs.get("model", "")
response_cost = kwargs.get("response_cost", 0.0) or 0
litellm_params = kwargs.get("litellm_params", {}) or {} litellm_params = kwargs.get("litellm_params", {}) or {}
_metadata = litellm_params.get("metadata", {}) _metadata = litellm_params.get("metadata", {})
proxy_server_request = litellm_params.get("proxy_server_request") or {} proxy_server_request = litellm_params.get("proxy_server_request") or {}
end_user_id = proxy_server_request.get("body", {}).get("user", None) end_user_id = proxy_server_request.get("body", {}).get("user", None)
user_id = litellm_params.get("metadata", {}).get("user_api_key_user_id", None) user_id = standard_logging_payload["metadata"]["user_api_key_user_id"]
user_api_key = litellm_params.get("metadata", {}).get("user_api_key", None) user_api_key = standard_logging_payload["metadata"]["user_api_key_hash"]
user_api_key_alias = litellm_params.get("metadata", {}).get( user_api_key_alias = standard_logging_payload["metadata"]["user_api_key_alias"]
"user_api_key_alias", None user_api_team = standard_logging_payload["metadata"]["user_api_key_team_id"]
) user_api_team_alias = standard_logging_payload["metadata"][
user_api_team = litellm_params.get("metadata", {}).get( "user_api_key_team_alias"
"user_api_key_team_id", None ]
) output_tokens = standard_logging_payload["completion_tokens"]
user_api_team_alias = litellm_params.get("metadata", {}).get( tokens_used = standard_logging_payload["total_tokens"]
"user_api_key_team_alias", None response_cost = standard_logging_payload["response_cost"]
)
_team_spend = litellm_params.get("metadata", {}).get( _team_spend = litellm_params.get("metadata", {}).get(
"user_api_key_team_spend", None "user_api_key_team_spend", None
) )
@ -312,12 +340,6 @@ class PrometheusLogger(CustomLogger):
_remaining_api_key_budget = safe_get_remaining_budget( _remaining_api_key_budget = safe_get_remaining_budget(
max_budget=_api_key_max_budget, spend=_api_key_spend max_budget=_api_key_max_budget, spend=_api_key_spend
) )
output_tokens = 1.0
if response_obj is not None:
tokens_used = response_obj.get("usage", {}).get("total_tokens", 0)
output_tokens = response_obj.get("usage", {}).get("completion_tokens", 0)
else:
tokens_used = 0
print_verbose( print_verbose(
f"inside track_prometheus_metrics, model {model}, response_cost {response_cost}, tokens_used {tokens_used}, end_user_id {end_user_id}, user_api_key {user_api_key}" f"inside track_prometheus_metrics, model {model}, response_cost {response_cost}, tokens_used {tokens_used}, end_user_id {end_user_id}, user_api_key {user_api_key}"
@ -358,7 +380,27 @@ class PrometheusLogger(CustomLogger):
user_api_team, user_api_team,
user_api_team_alias, user_api_team_alias,
user_id, user_id,
).inc(tokens_used) ).inc(standard_logging_payload["total_tokens"])
self.litellm_input_tokens_metric.labels(
end_user_id,
user_api_key,
user_api_key_alias,
model,
user_api_team,
user_api_team_alias,
user_id,
).inc(standard_logging_payload["prompt_tokens"])
self.litellm_output_tokens_metric.labels(
end_user_id,
user_api_key,
user_api_key_alias,
model,
user_api_team,
user_api_team_alias,
user_id,
).inc(standard_logging_payload["completion_tokens"])
self.litellm_remaining_team_budget_metric.labels( self.litellm_remaining_team_budget_metric.labels(
user_api_team, user_api_team_alias user_api_team, user_api_team_alias
@ -417,6 +459,7 @@ class PrometheusLogger(CustomLogger):
pass pass
async def async_log_failure_event(self, kwargs, response_obj, start_time, end_time): async def async_log_failure_event(self, kwargs, response_obj, start_time, end_time):
from litellm.types.utils import StandardLoggingPayload
verbose_logger.debug( verbose_logger.debug(
f"prometheus Logging - Enters failure logging function for kwargs {kwargs}" f"prometheus Logging - Enters failure logging function for kwargs {kwargs}"
@ -425,19 +468,18 @@ class PrometheusLogger(CustomLogger):
# unpack kwargs # unpack kwargs
model = kwargs.get("model", "") model = kwargs.get("model", "")
litellm_params = kwargs.get("litellm_params", {}) or {} litellm_params = kwargs.get("litellm_params", {}) or {}
standard_logging_payload: StandardLoggingPayload = kwargs.get(
"standard_logging_object", {}
)
proxy_server_request = litellm_params.get("proxy_server_request") or {} proxy_server_request = litellm_params.get("proxy_server_request") or {}
end_user_id = proxy_server_request.get("body", {}).get("user", None) end_user_id = proxy_server_request.get("body", {}).get("user", None)
user_id = litellm_params.get("metadata", {}).get("user_api_key_user_id", None) user_id = standard_logging_payload["metadata"]["user_api_key_user_id"]
user_api_key = litellm_params.get("metadata", {}).get("user_api_key", None) user_api_key = standard_logging_payload["metadata"]["user_api_key_hash"]
user_api_key_alias = litellm_params.get("metadata", {}).get( user_api_key_alias = standard_logging_payload["metadata"]["user_api_key_alias"]
"user_api_key_alias", None user_api_team = standard_logging_payload["metadata"]["user_api_key_team_id"]
) user_api_team_alias = standard_logging_payload["metadata"][
user_api_team = litellm_params.get("metadata", {}).get( "user_api_key_team_alias"
"user_api_key_team_id", None ]
)
user_api_team_alias = litellm_params.get("metadata", {}).get(
"user_api_key_team_alias", None
)
exception = kwargs.get("exception", None) exception = kwargs.get("exception", None)
try: try: