Distribute replication factor shardings more flexibly instead of their common prefixes. #965
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
copybara-service
bot
force-pushed
the
test_861661684
branch
2 times, most recently
from
ebda914 to
08d93bd
Compare
…r common prefixes. Only for cases that replicative factors are in all non-scalar tensors, for the moment. It may still produce suboptimal shardings for cases that some replicative factors are in all non-scalar tensors, and some are not. That is secondary because, this change still improves shardings for those cases, and those cases are not common among non-custom ops and sharding rules, for example, the replicative factor is only one one tensor, and the replicative factors for sort and concatenate are on all tensors. PiperOrigin-RevId: 862669962
copybara-service
bot
force-pushed
the
test_861661684
branch
from
08d93bd to
3ea912e
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Distribute replication factor shardings more flexibly instead of their common prefixes.
Only for cases that replicative factors are in all non-scalar tensors, for the moment.
It may still produce suboptimal shardings for cases that some replicative factors are in all non-scalar tensors, and some are not. That is secondary because, this change still improves shardings for those cases, and those cases are not common among non-custom ops and sharding rules, for example, the replicative factor is only one one tensor, and the replicative factors for sort and concatenate are on all tensors.