-
Notifications
You must be signed in to change notification settings - Fork 5.6k
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
add unittest for dataparallel bf16 O2 main_grad #57967
Conversation
你的PR提交成功,感谢你对开源项目的贡献! |
@@ -754,6 +754,8 @@ void EagerReducer::AddDistHook(size_t var_index) { | |||
auto &group = groups_[group_index]; | |||
auto &group_tensor = group.dense_tensors_[inside_group_index]; | |||
|
|||
if (!group_tensor.initialized()) return; |
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.
这个可以不要了,在这个pr里修复了:#57960 。可以rebase一下最新develop~
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.
done
4d5e23a
to
bfabcb3
Compare
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.
paddlenlp需要支持一下dp的main grad吧
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.
LGTM for set_tests_properties(test_dygraph_dataparallel_bf16 PROPERTIES TIMEOUT "200")
PR types
Others
PR changes
Others
Description
添加DP支持BF16 O2 main_grad的单测,验证方式和BF16 O1+白名单对比
PCard-70444