phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-24 10:14:26 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

(feat) cost: azure gpt + testing

Browse source
This commit is contained in:
ishaan-jaff 2023-11-23 14:20:48 -08:00
parent 90a9116a17
commit bb00595429
2 changed files with 28 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

28
litellm/tests/test_get_model_cost_map.py
View file

@ -55,4 +55,30 @@ def test_cost_ft_gpt_35():
assert cost == expected_cost
except Exception as e:
pytest.fail(f"Cost Calc failed for ft:gpt-3.5. Expected {expected_cost}, Calculated cost {cost}")
test_cost_ft_gpt_35()
# test_cost_ft_gpt_35()
def test_cost_azure_gpt_35():
try:
# this tests if litellm.completion_cost can calculate cost for azure/chatgpt-deployment-2 which maps to azure/gpt-3.5-turbo
# for this test we check if passing `model` to completion_cost overrides the completion cost
from litellm import ModelResponse, Choices, Message
from litellm.utils import Usage
resp = ModelResponse(
id='chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac',
choices=[Choices(finish_reason=None, index=0,
message=Message(content=' Sure! Here is a short poem about the sky:\n\nA canvas of blue, a', role='assistant'))],
model='chatGPT-deployment-LiteLLM-isAMAZING',
usage=Usage(prompt_tokens=21, completion_tokens=17, total_tokens=38)
)
cost = litellm.completion_cost(completion_response=resp, model="azure/gpt-3.5-turbo")
print("\n Calculated Cost for azure/gpt-3.5-turbo", cost)
input_cost = model_cost["azure/gpt-3.5-turbo"]["input_cost_per_token"]
output_cost = model_cost["azure/gpt-3.5-turbo"]["output_cost_per_token"]
expected_cost = (input_cost*resp.usage.prompt_tokens) + (output_cost*resp.usage.completion_tokens)
print("\n Excpected cost", expected_cost)
assert cost == expected_cost
except Exception as e:
pytest.fail(f"Cost Calc failed for azure/gpt-3.5-turbo. Expected {expected_cost}, Calculated cost {cost}")
# test_cost_azure_gpt_35()

2
litellm/utils.py
View file

@ -1678,7 +1678,7 @@ def completion_cost(
# get input/output tokens from completion_response
prompt_tokens = completion_response['usage']['prompt_tokens']
completion_tokens = completion_response['usage']['completion_tokens']
model = completion_response['model'] # get model from completion_response
model = model or completion_response['model'] # check if user passed an override for model, if it's none check completion_response['model']
else:
prompt_tokens = token_counter(model=model, text=prompt)
completion_tokens = token_counter(model=model, text=completion)