You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This will allow adding new schedulers locally without re-registering the default schedulers. The list gets stale when new default schedulers are added to TorchX
Detailed Proposal
Make it work the same way as named_resources (simply list all) or components (allow referencing via _0, _1, etc.)
The fix to simply merge/overwrite existing schedulers is trivial (publishing PR). The only question is whether we want to support the same _0/_1 behavior as for components here
clumsy
pushed a commit
to clumsy/torchx
that referenced
this issue
Apr 23, 2025
Description
Currently we can only have the default ones or the custom ones (from one of the packages): https://github.com/pytorch/torchx/blob/main/torchx/schedulers/__init__.py#L59
Motivation/Background
This will allow adding new schedulers locally without re-registering the default schedulers. The list gets stale when new default schedulers are added to TorchX
Detailed Proposal
Make it work the same way as named_resources (simply list all) or components (allow referencing via _0, _1, etc.)
Alternatives
Keep re-registering the default schedulers
Additional context/links
torchx/torchx/specs/finder.py
Line 322 in 4209218
The text was updated successfully, but these errors were encountered: