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 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
main features:
1.enhance expression of placement shard that shard the same tensor dim by many mesh dim by adding co_shard_order to support to merge many sharded tensor dim in reshape.
2.enhance reshard api to express that rearrange data before sharding tensor to support to reshard fused qkv in dist env.
main changes:
1.upgrade dims_mapping to be type of vector of vector
2.refactor nd_mesh reshard transform
3.add co_shard_order and split_factor in shard placement
4.add dims_mapping proxy to back compatible old spmd rule during transitional phase between dims_mapping of vector and new dims_mapping of vector of vector.
你的PR提交成功,感谢你对开源项目的贡献!
请关注后续CI自动化测试结果,详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.
liufengwei0103
changed the title
support to shard on the same tensor dim by many mesh dim, only dynami…
support to shard on the same tensor dim by many mesh dim, only dynamic graph
Jun 10, 2025
The results of the three example codes in the Description above also need to be explained through print or comments to make it easier for others to understand
The results of the three example codes in the Description above also need to be explained through print or comments to make it easier for others to understand
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
PR Category
Auto Parallel
PR Types
New features
Description
main features:
1.enhance expression of placement shard that shard the same tensor dim by many mesh dim by adding co_shard_order to support to merge many sharded tensor dim in reshape.
2.enhance reshard api to express that rearrange data before sharding tensor to support to reshard fused qkv in dist env.
main changes:
1.upgrade dims_mapping to be type of vector of vector
2.refactor nd_mesh reshard transform
3.add co_shard_order and split_factor in shard placement
4.add dims_mapping proxy to back compatible old spmd rule during transitional phase between dims_mapping of vector and new dims_mapping of vector of vector.
usage:
get a co_shard tensor
co shard in reshape
rearrange data before sharding
More use cases can be seen in the test cases.
Pcard-67164