Pydantic serializer warnings when dumping a completion response with logprobs

[kddubey]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

Dumping a completion response from a request with logprobs set raises this warning:

python3.11/site-packages/pydantic/main.py:308: UserWarning: Pydantic serializer warnings:
  Expected `int` but got `float` - serialized value may not be as expected
  Expected `int` but got `float` - serialized value may not be as expected
  Expected `int` but got `float` - serialized value may not be as expected
  return self.__pydantic_serializer__.to_python(

I confirmed that changing this type annotation to

    top_logprobs: Optional[List[Dict[str, float]]] = None

removes the warnings.

To Reproduce

1. Install openai>=1.0.0
2. In a completions API call, set logprobs>=1
3. Dump the response using .model_dump(), as advised in the migration guide

Code snippets

import openai

client = openai.OpenAI()

# Set logprobs >= 1
response = client.completions.create(
    model="curie", prompt="test", max_tokens=3, logprobs=1
)

response_as_dict = response.model_dump()
# raises a UserWarning

To ignore the warnings, the user currently has to do this:

import warnings

with warnings.catch_warnings():
    warnings.filterwarnings(
        "ignore",
        category=UserWarning,
        message="Pydantic serializer warnings",
    )
    response_as_dict = response.model_dump()

OS

macOS

Python version

Python v3.11.5

Library version

openai v1.1.2

[RobertCraigie]

Thanks for the report, this will be fixed in the next release. https://github.com/openai/openai-python/pull/738/files#diff-1d262046f5d37b8e4ee402833a96289d8264757a8f6b69bb843170aa295ca0cd

[RobertCraigie]

This is fixed in v1.2.0!