Skip to content

Remove THTensor::_dim, temporarily remove THTensor_nDimension. (#9835) #9895

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

Closed
wants to merge 1 commit into from

Conversation

gchanan
Copy link
Contributor

@gchanan gchanan commented Jul 26, 2018

Summary:
The primary goal here was to remove THTensor::_dim, which isn't part of the API moving forward.
Instead, we provide 3 options for getting the dimensionality (this is temporary although non-trivial to remove!):

nDimension                 corresponds to the "true" ATen dimension. TODO: implement.
nDimensionLegacyNoScalars  correpsonds to the ATen dimension, except scalars are viewed as 1-dimensional tensors.
nDimensionLegacyAll        corresponds to the ATen dimension, except scalars are viewed as 1-dimensional tensors
                           and tensors with a dimension of size zero are collapsed to 0-dimensional tensors.

So in this patch, nDimension -> nDimensionLegacyNoScalars and _dim/_nDimension goes to nDimensionLegacyAll.
These are just codemods.
Pull Request resolved: #9835

Reviewed By: ezyang

Differential Revision: D8999338

Pulled By: ezyang

…ch#9835)

Summary:
The primary goal here was to remove THTensor::_dim, which isn't part of the API moving forward.
Instead, we provide 3 options for getting the dimensionality (this is temporary although non-trivial to remove!):
```
nDimension                 corresponds to the "true" ATen dimension. TODO: implement.
nDimensionLegacyNoScalars  correpsonds to the ATen dimension, except scalars are viewed as 1-dimensional tensors.
nDimensionLegacyAll        corresponds to the ATen dimension, except scalars are viewed as 1-dimensional tensors
                           and tensors with a dimension of size zero are collapsed to 0-dimensional tensors.
```
So in this patch, nDimension -> nDimensionLegacyNoScalars and _dim/_nDimension goes to nDimensionLegacyAll.
These are just codemods.
Pull Request resolved: pytorch#9835

Reviewed By: ezyang

Differential Revision: D8999338

Pulled By: ezyang

fbshipit-source-id: 98e6c4f2f0a57034a8b3f318ce3f3ef734445fa6
zdevito pushed a commit to zdevito/ATen that referenced this pull request Jul 27, 2018
Summary:
Pull Request resolved: pytorch/pytorch#9895

The primary goal here was to remove THTensor::_dim, which isn't part of the API moving forward.
Instead, we provide 3 options for getting the dimensionality (this is temporary although non-trivial to remove!):
```
nDimension                 corresponds to the "true" ATen dimension. TODO: implement.
nDimensionLegacyNoScalars  correpsonds to the ATen dimension, except scalars are viewed as 1-dimensional tensors.
nDimensionLegacyAll        corresponds to the ATen dimension, except scalars are viewed as 1-dimensional tensors
                           and tensors with a dimension of size zero are collapsed to 0-dimensional tensors.
```
So in this patch, nDimension -> nDimensionLegacyNoScalars and _dim/_nDimension goes to nDimensionLegacyAll.
These are just codemods.
Pull Request resolved: pytorch/pytorch#9835

Reviewed By: ezyang

Differential Revision: D8999338

Pulled By: gchanan

fbshipit-source-id: a4d676ac728f6f36ca09604a41e888d545ae9311
jramseyer pushed a commit to jramseyer/pytorch that referenced this pull request Jul 30, 2018
…ch#9895)

Summary:
Pull Request resolved: pytorch#9895

The primary goal here was to remove THTensor::_dim, which isn't part of the API moving forward.
Instead, we provide 3 options for getting the dimensionality (this is temporary although non-trivial to remove!):
```
nDimension                 corresponds to the "true" ATen dimension. TODO: implement.
nDimensionLegacyNoScalars  correpsonds to the ATen dimension, except scalars are viewed as 1-dimensional tensors.
nDimensionLegacyAll        corresponds to the ATen dimension, except scalars are viewed as 1-dimensional tensors
                           and tensors with a dimension of size zero are collapsed to 0-dimensional tensors.
```
So in this patch, nDimension -> nDimensionLegacyNoScalars and _dim/_nDimension goes to nDimensionLegacyAll.
These are just codemods.
Pull Request resolved: pytorch#9835

Reviewed By: ezyang

Differential Revision: D8999338

Pulled By: gchanan

fbshipit-source-id: a4d676ac728f6f36ca09604a41e888d545ae9311
@alsrgv
Copy link

alsrgv commented Aug 2, 2018

@gchanan, what's the ETA for THTensor_nDimension() to be available again? This is breaking Horovod users who want to use master.

@gchanan
Copy link
Contributor Author

gchanan commented Aug 2, 2018

@alsrgv I can add it back ASAP -- though you can just use tensor->dim().

@gchanan
Copy link
Contributor Author

gchanan commented Aug 2, 2018

(unless you are just using the C API).

@alsrgv
Copy link

alsrgv commented Aug 2, 2018

We're still using C API, but we're planning to migrate to C++ soon.

goodlux pushed a commit to goodlux/pytorch that referenced this pull request Aug 15, 2018
…ch#9895)

Summary:
Pull Request resolved: pytorch#9895

The primary goal here was to remove THTensor::_dim, which isn't part of the API moving forward.
Instead, we provide 3 options for getting the dimensionality (this is temporary although non-trivial to remove!):
```
nDimension                 corresponds to the "true" ATen dimension. TODO: implement.
nDimensionLegacyNoScalars  correpsonds to the ATen dimension, except scalars are viewed as 1-dimensional tensors.
nDimensionLegacyAll        corresponds to the ATen dimension, except scalars are viewed as 1-dimensional tensors
                           and tensors with a dimension of size zero are collapsed to 0-dimensional tensors.
```
So in this patch, nDimension -> nDimensionLegacyNoScalars and _dim/_nDimension goes to nDimensionLegacyAll.
These are just codemods.
Pull Request resolved: pytorch#9835

Reviewed By: ezyang

Differential Revision: D8999338

Pulled By: gchanan

fbshipit-source-id: a4d676ac728f6f36ca09604a41e888d545ae9311
@ezyang ezyang added the merged label Jun 26, 2019
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