Terry Chen
Terry Chen
cannot repro the error from my side.
@asroy any plan on it?
Thank you for your suggestion. I think MiDaS Depth estimation model would indeed be valuable for ControlNet right? Regarding your question, is it the model you mentioned at this link:...
where did you apply torch_fix_seed?
what if put torch_fix_seed inside __main__?
We did fully test on Ampere GPUs ( A100), for most of kernels Turing GPUs(T4) should work well
you can try to disable use_mem_eff https://github.com/facebookincubator/AITemplate/blob/main/examples/05_stable_diffusion/src/modeling/attention.py#L72
> @terrychenism Any updates or future plans to validate T4 works well ? cc @ipiszy
@tchaton I don't have access to T4 gpu. Could you please run the model and localize the ops which are not supported on T4?
seems this PR https://github.com/facebookincubator/AITemplate/pull/126 breaks SD example.