Open rgxb2807 opened 8 months ago
@rgxb2807 i was hoping accelerate would just work lol
@rgxb2807 maybe you can raise an issue at the accelerate repository
@rgxb2807 i was hoping accelerate would just work lol
It does! and I'm grateful for it.
@rgxb2807 maybe you can raise an issue at the accelerate repository
I'll do some digging and raise an issue where appropriate.
Hi, thanks so much for the incredible repo.
I've been able to successfully train using a multi-gpu setup and accelerate. I'm wondering if it's possible to use Microsoft's DeepSpeed with the accelerate library.
When I enable Stage 1 or Stage 2 in the accelerate configuration I get the following error. Any ideas for workarounds?
RuntimeError: Tensor must have a storage_offset divisible by 2
I'm training SoundStream with the data from https://us.openslr.org/resources/12/dev-clean.tar.gz