PR Description updated to latest commit (ca53c81)

Just finalized the PR - It's ready for review now :)

Wow, thanks man @mspronesti Will review it ASAP

@shahules786 Thanks :)

This works very well, thanks @mspronesti!

Hi there @shahules786, any chance this PR gets reviewed? :)

Hi @mspronesti apologies for the late review. @jjmachan was testing this out. Will be reviewed soon.

Hi @mspronesti just finished testing and making some changes to avoid breaking changes.
I'm not sure if you tested it on any instruct models, but the changes made don't allow the use of instruct models.
Also if you have time can you remove the documentation changes that were made inorder to tackle the breaking change (with_openai)?

@jjmachan you can do the review now.

@shahules786 When I split this PR into two (the second part being #684) I didn't realize that completion models relied on the fact that this parameter needs to default to None not to bump into an IndexError from langchain (besides the point of being a mutable default).

I have appended one additional commit. Please try again :)

Regarding the docs, my suggestion would be to mark with_openai for deprecation (e.g., with a decorator) and discontinue it in the next major release. As such, I would advise showcasing the new usage in the docs, not the old one :)

Lastly, I find the name from_default ambiguous. I believe from_langchain was accurate and pythonic. What do you think?

hey @mspronesti - I agree from_defaults might be ambiguous so lets keep it as from_langchain itself as you first proposed.

as for deprecation I've added the code in this PR but let me think we have to remove it in this version

and for the completion bug - nice catch seems like it fixed the issue ❤️

Hi @jjmachan, thanks for the review :)

I wouldn't remove with_openai in this version, but in the next major release (0.2). I'd decorate it as deprecated now, suggesting from_lanchain as alternative. What do you think? :)

@jjmachan I've just added an extra commit to make the deprecation decorator more versatile and informative, and I've used it on with_openai to mark its deprecation as "pending" so that you can schedule that whenever convenient.

this is awesome @mspronesti - I love the depreciation util func - handy!
Once @shahules786 unblocks us we should be good to go and do a release later today as well

btw would love to hop on a call with you sometime if you are interested @mspronesti 🙂

@mspronesti Yes, that actually makes sense. Let's do that. I'll also add some more docs to help community use non-open ai models with it.