community: Additional parameters for OpenAIAssistantV2Runnable

[pm390]

Description: Added Additional parameters that could be useful for usage of OpenAIAssistantV2Runnable.

This change is thought to allow langchain users to set parameters that cannot be set using assistants UI (max_completion_tokens,max_prompt_tokens,parallel_tool_calls) and parameters that could be useful for experimenting like top_p and temperature.

This PR originated from the need of using parallel_tool_calls in langchain, this parameter is very important in openAI assistants because without this parameter set to False strict mode is not respected by OpenAI Assistants (https://platform.openai.com/docs/guides/function-calling#parallel-function-calling).

Note: Currently, if the model calls multiple functions in one turn then strict mode will be disabled for those calls.

Issue: None
Dependencies: openai

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)	Visit Preview		Jan 14, 2025 5:33pm

[pm390]

Actually...
I was just testing around the change, I noticed that the run_metadata parameter does not exist (and openAI doesn't have kwargs on their side) so invoking the assistant with the parameter set results in (if using async version):
TypeError: AsyncThreads.create_and_run() got an unexpected keyword argument 'run_metadata'

The correct parameter should be metadata.
I'll try get a minimal reproducer and open an Issue later this week.

[ccurme]

Actually... I was just testing around the change, I noticed that the run_metadata parameter does not exist (and openAI doesn't have kwargs on their side) so invoking the assistant with the parameter set results in (if using async version): TypeError: AsyncThreads.create_and_run() got an unexpected keyword argument 'run_metadata'

The correct parameter should be metadata. I'll try get a minimal reproducer and open an Issue later this week.

Thanks @pm390. From your perspective, is this PR good to go? i.e., the issue with run_metadata is separate? Or should we hold off?

[pm390]

@ccurme from my perspective we should treat the issue as a separate topic (adding a new functionality support should be separated from fixing an issue), so this PR is good to go.