search

neulab / prompt2model

prompt2model - Generate Deployable Models from Natural Language Instructions
Apache License 2.0
1.96k stars 177 forks source link

openai and litellm version confilicts #394

Open zhaochenyang20 opened 10 months ago

zhaochenyang20 commented 10 months ago

In the project.toml, we do not require a specific version of openai and litellm. If a user directly runs pip install ., then the newest version of openai may not support import openai.error. However, we use this in our api_tools.py. If we choose to use a lower version of openai, it is not supported by litellm. Thus we need to refactor our api_tools.py to support new version of openai and litellm.

viswavi commented 10 months ago

@zhaochenyang20: in pyproject.toml, we do specify specific versions of openai and litellm (https://github.com/neulab/prompt2model/blob/main/pyproject.toml)

zhaochenyang20 commented 10 months ago

Yeah. Make sense to me, but I personally think that we need to keep pace with the latest version of openai and iltellm. I gonna refactor this later.

saum7800 commented 8 months ago

I think this has become more important now. I was trying to use base model as gpt-4-turbo-preview but was getting litellm errors saying model could not be mapped correctly. I believe this is due to the mismatch. @zhaochenyang20 are you refactoring this?

viswavi commented 8 months ago

Hey I'm gonna take a crack at fixing this, hopefully this weekend! It's definitely something we can't ignore