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Outdated suggestions cannot be applied.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.
There were several issues with
Tensor.to()
inquanto
Tensor
subclasses.First, the
PackedTensor
andQbitsTensor
dispatch forTensor.to()
were not implemented.Second, when moving an entire
nn.Module
,pytorch
does not only callTensor.to()
on the weight/bias, but also tries to do a shallow copy into the existing weight/bias.Adding a proper dispatch at the function level to tag the
QTensor
class as incompatible with shallow copy solves the issue.