I've got some more changes related to distributed training on Ray I'm looking for some eyes on. https://github.com/flyteorg/flyte/pull/6232 • Adding support for custom ray head/worker affinity / tolerations for targeting GPU nodes