Skip to content

Finetuned OpenAI models cost calculation #11715 #12190

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Oct 24, 2023
Merged

Finetuned OpenAI models cost calculation #11715 #12190

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4bdb71b
Update fine tuned models dict
nirkopler Oct 24, 2023
1fe61e4
Update logic
nirkopler Oct 24, 2023
2116347
Update relevant tests
nirkopler Oct 24, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 17 additions & 5 deletions libs/langchain/langchain/callbacks/openai_info.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,10 +57,19 @@
"text-davinci-003": 0.02,
"text-davinci-002": 0.02,
"code-davinci-002": 0.02,
"ada-finetuned": 0.0016,
"babbage-finetuned": 0.0024,
"curie-finetuned": 0.012,
"davinci-finetuned": 0.12,
# Fine Tuned input
"babbage-002-finetuned": 0.0016,
"davinci-002-finetuned": 0.012,
"gpt-3.5-turbo-0613-finetuned": 0.012,
# Fine Tuned output
"babbage-002-finetuned-completion": 0.0016,
"davinci-002-finetuned-completion": 0.012,
"gpt-3.5-turbo-0613-finetuned-completion": 0.016,
# Legacy fine-tuned models
"ada-finetuned-legacy": 0.0016,
"babbage-finetuned-legacy": 0.0024,
"curie-finetuned-legacy": 0.012,
"davinci-finetuned-legacy": 0.12,
}


Expand All @@ -82,11 +91,14 @@ def standardize_model_name(
"""
model_name = model_name.lower()
if "ft-" in model_name:
return model_name.split(":")[0] + "-finetuned"
return model_name.split(":")[0] + "-finetuned-legacy"
if "ft:" in model_name:
return model_name.split(":")[1] + "-finetuned"
Comment on lines +95 to +96
Copy link
Contributor

@mspronesti mspronesti Oct 24, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this work on Azure OpenAI? It looks like Azure OpenAI fine-tuned models have the following structure
<model_name>.ft-<job_id> (reference), which would go into the previous if case but without doing the right thing I fear :/

Copy link
Contributor Author

@nirkopler nirkopler Oct 25, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

youre right, Ill open a new PR about the azure fix, Ill tag you 💪

elif is_completion and (
model_name.startswith("gpt-4")
or model_name.startswith("gpt-3.5")
or model_name.startswith("gpt-35")
or ("finetuned" in model_name and "legacy" not in model_name)
):
return model_name + "-completion"
else:
Expand Down
18 changes: 16 additions & 2 deletions libs/langchain/tests/unit_tests/callbacks/test_openai_info.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,7 +49,21 @@ def test_on_llm_end_custom_model(handler: OpenAICallbackHandler) -> None:
assert handler.total_cost == 0


def test_on_llm_end_finetuned_model(handler: OpenAICallbackHandler) -> None:
@pytest.mark.parametrize(
"model_name",
[
"ada:ft-your-org:custom-model-name-2022-02-15-04-21-04",
"babbage:ft-your-org:custom-model-name-2022-02-15-04-21-04",
"curie:ft-your-org:custom-model-name-2022-02-15-04-21-04",
"davinci:ft-your-org:custom-model-name-2022-02-15-04-21-04",
"ft:babbage-002:your-org:custom-model-name:1abcdefg",
"ft:davinci-002:your-org:custom-model-name:1abcdefg",
"ft:gpt-3.5-turbo-0613:your-org:custom-model-name:1abcdefg",
],
)
def test_on_llm_end_finetuned_model(
handler: OpenAICallbackHandler, model_name: str
) -> None:
response = LLMResult(
generations=[],
llm_output={
Expand All @@ -58,7 +72,7 @@ def test_on_llm_end_finetuned_model(handler: OpenAICallbackHandler) -> None:
"completion_tokens": 1,
"total_tokens": 3,
},
"model_name": "ada:ft-your-org:custom-model-name-2022-02-15-04-21-04",
"model_name": model_name,
},
)
handler.on_llm_end(response)
Expand Down