Sebastián Ramírez

I would think now there would be fewer warnings as the `autocompletion` parameter is now undeprecated. :sweat_smile: And the other `shell_complete` parameter was never documented, and the implementation was not...

@FurkanGozukara please share a self-contained [minimal self-contained example](https://stackoverflow.com/help/minimal-reproducible-example) that replicates the issue so that I can copy-paste it, run it, and see your error. I just tried with the getting-started...

Thanks for the minimal app! I'm being able to make it break with that, so now I can debug it. --- I guess Gradio and vllm are using some of...

About Gradio, I debugged it and reported here: https://github.com/gradio-app/gradio/issues/9278#issuecomment-2333875064. It's something that should be refactored in Gradio (the current code is doing something that shouldn't be done), they are aware...

@drikster80 can you provide a self-contained [minimal, reproducible example](https://stackoverflow.com/help/minimal-reproducible-example) of how vllm is not working? I'm trying it here with the quickstart and it all seems to work.

@cobanov about https://github.com/bmaltais/kohya_ss/issues/2783, that repo seems to be a Gradio app, so that would be fixed once they fix and release that in Gradio.

@drikster80 interesting, I wouldn't think those lines would be a problem as those are parameters for functions, not Pydantic models used for inputs in an API. :thinking:

@drikster80 > I believe I was able to find a solution to this. It is related to https://github.com/openai/openai-python/issues/1454 and https://github.com/fastapi/fastapi/discussions/11952 I would think it's not related, those two links are...

Thanks @drikster80 and @pachewise! Checking @pachewise's code in that comment, I was able to reduce the error reproduction to only using Pydantic's `TypeAdapter`. I commented with it here: https://github.com/vllm-project/vllm/issues/8212#issuecomment-2342848855 It...

Thanks @dzhulgakov, I'll look into it.