Permission Error: [Errno 13] Permission denied

[sTiKyt]

[chrismelbourne]

Are you on a Mac? If so you shouldn't be running CUDA, or even bitsandbytes (which uses CUDA).

[sTiKyt]

Are you on a Mac? If so you shouldn't be running CUDA, or even bitsandbytes (which uses CUDA).

No, i'm on Arch Linux

And this one works just fine:
https://github.com/easydiffusion/easydiffusion

Just in case in matters, but i want to train some LoRA models myself

[TehNomad]

I have the same error with Arch Linux and I'm also using LightDM. It only happens to me with the Adam8Wbit optimizer, DAdaptAdam works fine for me.

[sTiKyt]

I have the same error with Arch Linux and I'm also using LightDM. It only happens to me with the Adam8Wbit optimizer, DAdaptAdam works fine for me.

Dude, thanks a lot, you have no idea how much you just helped me xD

I'm just experimenting and don't really understand most of the things i'm doing so i would never try changing optimizer to fix that crap, now it seems to be training as expected

[sTiKyt]

I have the same error with Arch Linux and I'm also using LightDM. It only happens to me with the Adam8Wbit optimizer, DAdaptAdam works fine for me.

Dude, thanks a lot, you have no idea how much you just helped me xD

I'm just experimenting and don't really understand most of the things i'm doing so i would never try changing optimizer to fix that crap, now it seems to be training as expected

Could you explain what does optimizer even do and where to look while picking one?

[TehNomad]

The optimizer is the algorithm that adjusts the parameters of your model during training to improve its accuracy. Check out this document for how to use dAdapatation:
https://rentry.org/LazyDAdaptationGuide