Initialize gradient of uninitialized parameter with default dtype when initializer is callable #5064
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Outdated suggestions cannot be applied.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.
This PR follows up #4510.
The gradient of an uninitialized parameter was always initialized with NaN of
numpy.float32
when its initializer was a callable, in contrast to its data that was initialized with the default dtype. This caused mismatch of dtypes between the data and the gradient.This PR fixes it to use the default dtype to initialize the gradient in such case so that the data and the gradient have the same dtype.