search

ray-project / ray

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
https://ray.io
Apache License 2.0
33.11k stars 5.6k forks source link

[ml][Improvement] Improve messages to be “rank0, rank1” actors etc. #23310

Open richardliaw opened 2 years ago

richardliaw commented 2 years ago

The node ip and process doesn’t really make a lot of sense when you are debugging distributed training.

  1. Is not hard blocker
  2. Is persistent
  3. Is easy to work around
ericl commented 2 years ago

This just means defining the __repr__ of the actors right?

richardliaw commented 2 years ago

should be, right

On Thu, Mar 17, 2022 at 8:02 PM Eric Liang @.***> wrote:

This just means defining the repr of the actors right?

— Reply to this email directly, view it on GitHub https://github.com/ray-project/ray/issues/23310#issuecomment-1071980957, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABCRZZMDSRDLEXO5RL46VDLVAPW3PANCNFSM5RANAC3Q . You are receiving this because you authored the thread.Message ID: @.***>

scottsun94 commented 1 year ago

Is this only for actors created by ray train or any actor?

amogkam commented 1 year ago

Just for actors created by Ray Train.