-
Notifications
You must be signed in to change notification settings - Fork 4.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[tensor] redistribute among different process groups #1247
Conversation
if pg is not None and pg != self.get_process_group(): | ||
print('here _redistribute') | ||
# if the pg is not equal, convert the current tensor to replicated | ||
self._redistribute(ReplicaSpec()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think the updated operand, redistribute
, always used for non-model data in training.
It natrual that we need to keep redistribute
as an autograd operand.
But here, _redistribute
is not capable being used as an autograd operand.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I see. I will discuss with you offline.
I hope the ColoTensor can be used independently, we don't need to assume it's only used for training.
No description provided.