[BC] Allow only bool tensors as mask in masked_select
#96112
Conversation
`byte` support was marked as deprecated in 1.13, so it's fine to remove it in 2.1
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/96112
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f5152ec: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
malfet
added
release notes: python_frontend
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
| @@ -1899,12 +1894,12 @@ static Tensor & masked_select_out_impl_cpu(Tensor & result, const Tensor & self, | |||
| auto mask_long_data = mask_long.data_ptr<int64_t>(); | |||
| auto mask_prefix_sum_data = mask_prefix_sum.data_ptr<int64_t>(); | |||
| // TODO: Here can only use std::partial_sum for C++14, | |||
| // use std::exclusive_scan when PyTorch upgrades to C++17, which have better peformance. | |||
| // use std::exclusive_scan when PyTorch upgrades to C++17, which have better performance. | |||
| // std::exclusive_scan(mask_long_data, mask_long_data + mask_long.numel(), mask_prefix_sum_data, 0); | |||
There was a problem hiding this comment.
@malfet can you swap out the comments to std::exclusive_scan now that we are in C++17?
There was a problem hiding this comment.
I'll do it (and few other changes) in followup PR (don't want to muddy this one with unrelated functional changes, which also can cause reverts due to ancient internal compilers or something of the natrue)
Merge failedReason: 1 jobs have failed, first few of them are: trunk / macos-12-py3-arm64 / test (default, 1, 2, macos-m1-12) Details for Dev Infra teamRaised by workflow job |
Do not test for `torch.uint8` and expect `torch.masked_select` to raise an assert
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 1 mandatory check(s) failed. The first few are: Dig deeper by viewing the failures on hud |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
) `byte` support was marked as deprecated in 1.8, so it's fine to remove this in 2.1 (or even 2.0) Deprecation warning was added by pytorch#22261 Also, fix bunch of syntactic errors in comments Pull Request resolved: pytorch#96112 Approved by: https://github.com/ezyang
`byte` support was marked as deprecated in 1.8, so it's fine to remove this in 2.1 (or even 2.0) Deprecation warning was added by pytorch/pytorch#22261 Also, fix bunch of syntactic errors in comments Pull Request resolved: pytorch/pytorch#96112 Approved by: https://github.com/ezyang
`byte` support was marked as deprecated in 1.8, so it's fine to remove this in 2.1 (or even 2.0) Deprecation warning was added by pytorch/pytorch#22261 Also, fix bunch of syntactic errors in comments Pull Request resolved: pytorch/pytorch#96112 Approved by: https://github.com/ezyang
__What?__ Per discussion at #94634, deprecate `masked_fill` with non-bool masks. Deprecation warnings were previously added by #22261, but not for Apple MPS. I can revert the MPS changes if deprecation warnings are wanted first tho. See also #96112. Fixes #85063 and #89320. __Further Development?__ - Fixed the mask dtype checking for the cuda dispatch for `masked_fill` in `aten/src/ATen/native/cuda/Indexing.cu` Pull Request resolved: #96594 Approved by: https://github.com/malfet, https://github.com/ngimel
) `byte` support was marked as deprecated in 1.8, so it's fine to remove this in 2.1 (or even 2.0) Deprecation warning was added by pytorch#22261 Also, fix bunch of syntactic errors in comments Pull Request resolved: pytorch#96112 Approved by: https://github.com/ezyang
bytesupport was marked as deprecated in 1.8, so it's fine to remove this in 2.1 (or even 2.0)Deprecation warning was added by #22261
Also, fix bunch of syntactic errors in comments
cc @albanD