What you would like to be added?
Currently there is a Kubernetes backend for full remote distributed training and SubProcess backend for Local execution.
The proposal is to add an additional backend for Docker, also to be used in a Local context.
Why is this needed?
This bridges the gap between the current subprocess backend and a full kubernetes backend by facilitating training in a containerised environment and across multiple containers.
Love this feature?
Give it a 👍 We prioritize the features with most 👍