Jeff Rasley

Hi @zarzen and @szhengac, thanks for reaching out with this proposal! We’re discussing and will respond later next week.

@zarzen, I just sent you an email please let me know if you don't receive it for some reason :)

@lw3259111, @jiacheng-ye, @james-yw, @lixinliu1995: thank you all for using DeepSpeed, please re-open if v0.8.3 does not fix your issue please re-open the issue

@Wenhan-Tan we do not currently support BF16 directly with deepspeed inference. @molly-smith can we add an assertion in the ds-inference config to gracefully error out if someone tries this case?

closing in favor of https://github.com/microsoft/DeepSpeed/pull/2995

Thank you @chunhualiao, we’re working on adding an option for automatic feature selection that should help here. We realize it’s not always easy to determine which arguments should be used...

If you want to add arguments to the training such as the ones you list above (e.g., --gradient_checkpointing) you'll need to add them after `main.py` in the script for example:...

> can it run on the macbook pro with m1 pro cpu? Sorry, but we don't currently support M1. I know torch now supports it but the biggest blocker here...

> @tjruwase we plan to add a new workflow to validate installation for CPU and run some inference related test. If there is anything we should know before add new...

A small update on this thread @stas00, @awan-10 just discovered an issue with the latest version of FusedAdam from apex that has to do with this PR: https://github.com/NVIDIA/apex/pull/1560. A snippet...