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.

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

fix adam with multi dim #39

Merged
merged 11 commits into from
Jun 28, 2022
Merged

fix adam with multi dim #39

merged 11 commits into from
Jun 28, 2022

Conversation

danleifeng
Copy link

@danleifeng danleifeng commented Jun 20, 2022

PR types

Bug fixes

PR changes

Others

Describe

  1. 支持多种维度
  2. 修复 embed 初始化在adam下 非0初始化 (zero_init)
  3. 修复adam和 shared_adam update初始化时 不必要的 embed_beta1 embed_beta2赋值

@danleifeng danleifeng changed the title fix adam fix adam with multi dim Jun 27, 2022
@xuewujiao xuewujiao merged commit 7263442 into xuewujiao:gpugraph Jun 28, 2022
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Nov 24, 2022
…addle#46342)

* add extra attr property set

* add type_info for all context

* add onednn context to all context

* fix context compile error

* simplify conv kernel args

* pass runtime attr into dev_ctx

* fix marco error

* clear conv_grad_kernel extra args

* merge conv_grad_grad into conv_grad

* clear conv2d_grad_grad extra attrs

* clear yaml and eager extra attr

* fix conv1d error

* change to thread local

* fix npu compile failed

* try to fix windows compile failed

* add conv2d onednn phi kernel

* fix ci bugs (xuewujiao#36)

* fix compile bugs (xuewujiao#38)

* fix extra input transform bug (xuewujiao#39)

* support dynamic created attr (xuewujiao#40)

* reset extra info gen code

* rm conv_grad_grad kernel

* reimpl pass attr adapting

* add int attr support

* remove vector inputnames creating

* fix map at error

* Update paddle/phi/kernels/onednn/conv_grad_kernel.cc

Co-authored-by: Sławomir Siwek <[email protected]>

* remove useless extra attrs

* replace mkldnn_engine by onednn_engine

Co-authored-by: YuanRisheng <[email protected]>
Co-authored-by: Sławomir Siwek <[email protected]>
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Nov 24, 2022
* add extra attr property set

* add type_info for all context

* add onednn context to all context

* fix context compile error

* simplify conv kernel args

* pass runtime attr into dev_ctx

* fix marco error

* clear conv_grad_kernel extra args

* merge conv_grad_grad into conv_grad

* clear conv2d_grad_grad extra attrs

* remove redundant imports

* migrate softmax

* clear yaml and eager extra attr

* fix conv1d error

* change to thread local

* fix npu compile failed

* try to fix windows compile failed

* add conv2d onednn phi kernel

* fix ci bugs (xuewujiao#36)

* fix compile bugs (xuewujiao#38)

* fix extra input transform bug (xuewujiao#39)

* support dynamic created attr (xuewujiao#40)

* reset extra info gen code

* rm conv_grad_grad kernel

* reimpl pass attr adapting

* add int attr support

* remove vector inputnames creating

* merge dev

* fix map at error

* adjust attribute

* adapt funcs to PHI

Co-authored-by: Chen Weihang <[email protected]>
Co-authored-by: YuanRisheng <[email protected]>
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Nov 24, 2022
* add extra attr property set

* add type_info for all context

* add onednn context to all context

* fix context compile error

* simplify conv kernel args

* pass runtime attr into dev_ctx

* fix marco error

* clear conv_grad_kernel extra args

* merge conv_grad_grad into conv_grad

* clear conv2d_grad_grad extra attrs

* remove redundant imports

* migrate softmax

* clear yaml and eager extra attr

* fix conv1d error

* change to thread local

* fix npu compile failed

* try to fix windows compile failed

* add conv2d onednn phi kernel

* fix ci bugs (xuewujiao#36)

* fix compile bugs (xuewujiao#38)

* fix extra input transform bug (xuewujiao#39)

* support dynamic created attr (xuewujiao#40)

* reset extra info gen code

* rm conv_grad_grad kernel

* reimpl pass attr adapting

* add int attr support

* remove vector inputnames creating

* merge dev

* fix map at error

* adjust attribute

* adapt funcs to PHI

* init

* adjust imports

* support postops

* format codeblocks

* revert changes to softmax

Co-authored-by: Chen Weihang <[email protected]>
Co-authored-by: YuanRisheng <[email protected]>
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Nov 24, 2022
* add extra attr property set

* add type_info for all context

* add onednn context to all context

* fix context compile error

* simplify conv kernel args

* pass runtime attr into dev_ctx

* fix marco error

* clear conv_grad_kernel extra args

* merge conv_grad_grad into conv_grad

* clear conv2d_grad_grad extra attrs

* clear yaml and eager extra attr

* fix conv1d error

* change to thread local

* fix npu compile failed

* try to fix windows compile failed

* add conv2d onednn phi kernel

* fix ci bugs (xuewujiao#36)

* fix compile bugs (xuewujiao#38)

* fix extra input transform bug (xuewujiao#39)

* support dynamic created attr (xuewujiao#40)

* reset extra info gen code

* rm conv_grad_grad kernel

* reimpl pass attr adapting

* add int attr support

* remove vector inputnames creating

* fix map at error

* Update paddle/phi/kernels/onednn/conv_grad_kernel.cc

Co-authored-by: Sławomir Siwek <[email protected]>

* remove useless extra attrs

* replace mkldnn_engine by onednn_engine

* Migrate pool+grad to PHI

* Update paddle/fluid/operators/mkldnn/test_mkldnn_op_nhwc.cc

Co-authored-by: Sławomir Siwek <[email protected]>

* Update paddle/phi/kernels/onednn/pool_grad_kernel.cc

Co-authored-by: Sławomir Siwek <[email protected]>

* Update paddle/phi/kernels/onednn/pool_kernel.cc

Co-authored-by: Sławomir Siwek <[email protected]>

Co-authored-by: Chen Weihang <[email protected]>
Co-authored-by: YuanRisheng <[email protected]>
Co-authored-by: Chen Weihang <[email protected]>
Co-authored-by: Sławomir Siwek <[email protected]>
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Dec 17, 2022
* fix adam with multi dim; test=develop
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants