LGTM! Pinging @sayakpaul, @yiyixuxu @DN6 for a quick peek since the PR also touches the test scripts :)
Just a single comment but rest looks good to me.
896 | 896 | images = [] | |
897 | 897 | if args.validation_prompts is not None: | |
898 | 898 | logger.info("Running inference for collecting generated images...") | |
899 | pipeline = pipeline.to(accelerator.device) |
Why is this going away?
it called enable_model_cpu_offload()
right after
thanks!
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.
Thanks for merging!
Login to write a write a comment.
Fixes CPU offloading usage.
@stevhliu