Refactored GradNodeAccumulation data structure and behaviour #39526
Conversation
This file contains 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
|
Thanks for your contribution! |
jim19930609
force-pushed
the
fix_accumulation_node
branch
from
352c972 to
aab36af
Compare
… fix_accumulation_node
… fix_accumulation_node
jim19930609
force-pushed
the
fix_accumulation_node
branch
from
490a1dc to
200806d
Compare
… fix_accumulation_node
… fix_accumulation_node
jim19930609
force-pushed
the
fix_accumulation_node
branch
from
051f10a to
d342316
Compare
… fix_accumulation_node
… fix_accumulation_node
chenwhql
approved these changes
Feb 24, 2022
| std::dynamic_pointer_cast<egr::GradNodeAccumulation>(grad_node); | ||
| grad = accumulation_grad_node->Grad(); | ||
| grad = egr::EagerUtils::mutable_grad(self->tensor); | ||
| PADDLE_ENFORCE(grad != nullptr, |
There was a problem hiding this comment.
推荐直接用PADDLE_ENFORCE_NOT_NULL
| grad = egr::EagerUtils::mutable_grad(self->tensor); | ||
| PADDLE_ENFORCE(grad != nullptr, | ||
| paddle::platform::errors::Fatal( | ||
| "Detected NULL grad" |
There was a problem hiding this comment.
Detected NULL grad后面需要标点符号断下句吗？报错结尾建议用句点结束
JiabinYang
approved these changes
Feb 24, 2022
| @@ -52,9 +52,15 @@ void RegisterReduceHookForTensor(const paddle::experimental::Tensor& tensor, | |||
| } | |||
| } | |||
|
|
|||
| void RetainGradForTensor(const paddle::experimental::Tensor& tensor) { | |||
| // TODO(jiabin): Support More Tensor type here | |||
| static void RetainGradForRegularNode( | |||
There was a problem hiding this comment.
merge this with RegisterGradientHookForTensor
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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 types
New features
PR changes
Others
Describe
We used to register RetainGrad hook for leaf tensor so as to update grad tensor's value. After this patch, we asked GradNodeAccumulation to track a weak_ptr of grad_tensor, the value of which will get updated upon calling GradNodeAccumulation::operator(), instead of having to register a RetainGrad hook.