Skip to content
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.

Sign up for GitHub

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

Jump to bottom

two-phase training for ps #40762

Merged
merged 53 commits into from
Mar 23, 2022
Merged

two-phase training for ps #40762

merged 53 commits into from
Mar 23, 2022

Conversation

zhaocaibei123
Copy link
Contributor

@zhaocaibei123 zhaocaibei123 commented Mar 21, 2022 •
edited
Loading

PR types

New features

PR changes

Others

Describe

Paddle distributed framework support two-phase training for ps async mode now, and we will provide two-phase training demo in PaddleRec later.

  1. move BrpcPSClient from Communicator to FleetWrapper, remove Communicator for ps async mode
  2. add DownpourLiteWorker to support more function than HogwilderWorker, like metric and others
  3. add SparseAccessor for common embedding, and modify CommonCtrAccessor for embedding with show/click
  4. make new compile-time refactor for programs takes into effect, and fix some bugs

esythan and others added 30 commits February 11, 2022 07:44
@esythan
fix benchmark and communicator config
56a9e6d
@esythan
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
cce6594 
… new_one_ps
@esythan
fix bugs of the_one_ps
14cb9f4
@esythan
merge develop
fe81323
@esythan
multi program and fix bug in optimizer
972ddb8
@esythan
multi program in the_one_ps
96af1d4
@esythan
public commcontext
928032b
@esythan
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
d7c95de 
… ps_optimizer
@esythan
ps optimizer multi programs
c7e9a6e
@zhaocaibei123
cvm & datanorm backend
398e47f
@zhaocaibei123
fix dim
93d494a
@zhaocaibei123
fix unittest
61a2cb2
@zhaocaibei123
fix
0017e1a
@esythan
merge develop
7dcd805
@esythan
the one ps merge
63fa39b
@zhaocaibei123
remove comm
bedfa6e
@zhaocaibei123
add DownpourLiteWorker
7de4ea4
@esythan
Merge branch 'trainer_desc' into cvm_backend
9f74ad8
@esythan
Merge pull request #11 from zhaocaibei123/cvm_backend
cb0bef1 
Cvm backend
@esythan
all
9ba31b4
@zhaocaibei123
fix
83653ae
@esythan
Merge pull request #12 from zhaocaibei123/cvm_backend
242cfcc 
fix
@esythan
fix
9cc1686
@esythan
device worker downpour lite
297f8d3
@zhaocaibei123
fix
a248d09
@esythan
Merge pull request #13 from zhaocaibei123/cvm_backend
0398658 
fix
@esythan
fix bug in global shuffle
c76cf94
@esythan
save inference model
c39ca28
@zhaocaibei123
fix & add log
b490ec9
@zhaocaibei123
Merge branch 'trainer_desc' of https://github.com/esythan/Paddle into…
355ca08 
… cvm_test
@paddle-bot-old
Copy link

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.

@paddle-bot-old
Copy link

paddle-bot-old bot commented Mar 21, 2022 •
edited
Loading

✅ This PR's description meets the template requirements!
Please wait for other CI results.

@zhaocaibei123 zhaocaibei123 changed the title Cvm 20220321 two-phase training Mar 22, 2022
yaoxuefeng6
yaoxuefeng6 previously approved these changes Mar 22, 2022
View reviewed changes
Copy link
Contributor

@yaoxuefeng6 yaoxuefeng6 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

Shixiaowei02
Shixiaowei02 approved these changes Mar 23, 2022
View reviewed changes
Copy link
Contributor

@Shixiaowei02 Shixiaowei02 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

approve for the change of distributed_push_sparse_op

@zhaocaibei123 zhaocaibei123 changed the title two-phase training two-phase training for ps Mar 23, 2022
XiaoguangHu01
XiaoguangHu01 approved these changes Mar 23, 2022
View reviewed changes
Copy link
Contributor

@XiaoguangHu01 XiaoguangHu01 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@yaoxuefeng6 yaoxuefeng6 merged commit b1a4668 into PaddlePaddle:develop Mar 23, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@XiaoguangHu01 XiaoguangHu01 XiaoguangHu01 approved these changes

@Shixiaowei02 Shixiaowei02 Shixiaowei02 approved these changes

@fuyinno4 fuyinno4 fuyinno4 approved these changes

@PangHua PangHua PangHua approved these changes

@yaoxuefeng6 yaoxuefeng6 yaoxuefeng6 left review comments

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
7 participants
@zhaocaibei123 @PangHua @fuyinno4 @Shixiaowei02 @XiaoguangHu01 @yaoxuefeng6 @esythan