Benjamin Bossan

Not sure why the upgrade was reverted, but `build_pr_documentation` starts [failing now for PEFT](https://github.com/huggingface/peft/actions/runs/12812305995/job/35723635733?pr=2287#logs). Will this be addressed soon or do we need to change something on the PEFT side?

@pengjunfeng11 Thanks for raising this issue. If you are interested, please feel free to create a PR to update the docs. If you believe the [examples](https://github.com/huggingface/peft/tree/main/examples/sft) also need to be...

@pengjunfeng11 Are you still interested in working on this?

Thanks for investigating this further. I wouldn't delete the accelerate content. Although it is true that DeepSpeed can be used without accelerate, in practice many people do use it with...

> I will draw a new section just for `deepspeed` and `PEFT `, and keep `accelearate` part. Thanks. > Or do you think there are any areas that could be...

~CI needs #1070 first, as torch 2.5 is not available for Python 3.8.~

Thanks for creating this PR. Just to ensure that I got things right: Let's assume we have a negative score, say, -10, and a threshold of 0.1. Then, with the...

> Yes, you described the issue exactly as I see it. Great, thanks for confirming. > Regarding the test: I am not as familiar with code, but I think I...

@MaxBalmus gentle ping :)

I agree that the information is a bit sparse. Could you expand on what exactly you would like to see? What is the workaround for that you mentioned, do you...