Remove transpose from reshape builder#22901
Closed
shraiysh wants to merge 3 commits intoopenxla:mainfrom
Closed
Conversation
49b54f9 to
eb35b77
Compare
Contributor
Author
|
The tensorflow CI failure is because of no storage on device, and is not related to this PR. |
Member
|
If this is changed, the operation semantics should be changed as well: In general I think it is probably a good idea to change this, as for stablehlo we already don't allow the implicit transpose: |
Contributor
Author
Can you tell me where the source for this is? |
Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required.
eb35b77 to
b250412
Compare
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 25, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 26, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 26, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 26, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 26, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
Contributor
Author
|
Are there internal failures with this PR? Let me know if there is anything I can do to help merge this. |
Member
|
There were internal adjustments needed. I did the last ones yesterday, but I need approval for them and unfortunately so far I didn't get it. Hopefully I can merge it either later today or tomorrow. |
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR #22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc31 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cf by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b250412 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=#22901 from shraiysh:fix_reshape_and_shape_inference b250412 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#22901 from shraiysh:fix_reshape_and_shape_inference b25041292522834a544b2c7c463909fc576c4ba9 PiperOrigin-RevId: 730737886
copybara-service Bot
pushed a commit
to tensorflow/tensorflow
that referenced
this pull request
Feb 28, 2025
Imported from GitHub PR openxla/xla#22901 Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which permutation of the layoyt of the shape. 2. `new_sizes` which are the new dimensions of the shape. This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required. Copybara import of the project: -- b44dc317188053cd51b7af4e90598ba357ba1197 by Shraiysh Vaishay <svaishay@nvidia.com>: Remove transpose from reshape builder Reshape builder unnecessarily accepts two arguments: 1. `dimensions` which is the "layout" of the shape. 2. `new_sizes` which is the "dimensions" of the shape. This is confusing, and the we can simply remove the support for layout - as it is just a transpose and such a transpose can be easily constructed wherever required. -- a7100cfc2185f5b8fd9dcf6dfa52c221b195021c by Shraiysh Vaishay <svaishay@nvidia.com>: Fix unrelated clang-format issues in xla/python/ops.cc -- b25041292522834a544b2c7c463909fc576c4ba9 by Shraiysh Vaishay <svaishay@nvidia.com>: Fix semantics doc Merging this change closes #22901 PiperOrigin-RevId: 732082186
tengyifei
added a commit
to pytorch/xla
that referenced
this pull request
Mar 5, 2025
I need to rewrite reshape usage according to openxla/xla#22901 and handle some breaking xla::ifrt changes.
tengyifei
added a commit
to pytorch/xla
that referenced
this pull request
Mar 5, 2025
I need to rewrite reshape usage according to openxla/xla#22901 and handle some breaking xla::ifrt changes. I also need to force HLO printing to use the Default (long) format due to https://github.com/openxla/xla/pull/22614/files
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Reshape builder unnecessarily accepts two arguments:
dimensionswhich permutation of the layoyt of the shape.new_sizeswhich are the new dimensions of the shape.This is confusing, and the we can simply remove the support for permutation of layout - as it is just a transpose and such a transpose can be easily constructed wherever required.