The `usort` config in `pyproject.toml` has no effect due to a typo. Fixing the typo make `usort` do more and generate the changes in the PR. Except `pyproject.toml`, all changes are generated by `lintrunner -a --take UFMT --all-files`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/127126
Approved by: https://github.com/kit1980
The `usort` config in `pyproject.toml` has no effect due to a typo. Fixing the typo make `usort` do more and generate the changes in the PR. Except `pyproject.toml`, all changes are generated by `lintrunner -a --take UFMT --all-files`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/127126
Approved by: https://github.com/kit1980
ghstack dependencies: #127122, #127123, #127124, #127125
Summary:
caffe2/test:torch_cuda - test_advanced_indexing_assignment_lazy (test_view_ops.TestViewOpsLAZY)
RuntimeError: TorchScript backend not yet supported in FBCODE/OVRSOURCE builds
File "/usr/local/fbcode/platform010/lib/python3.8/unittest/suite.py", line 163, in _handleClassSetUp
setUpClass()
File "/re_cwd/fbcode/buck-out/opt/gen/caffe2/test/torch_cuda#binary,link-tree/torch/testing/_internal/common_device_type.py", line 506, in setUpClass
torch._lazy.ts_backend.init()
File "/re_cwd/fbcode/buck-out/opt/gen/caffe2/test/torch_cuda#binary,link-tree/torch/_lazy/ts_backend.py", line 6, in init
torch._C._lazy_ts_backend._init()
Test Plan: Rely on CI.
Differential Revision: D41170545
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88786
Approved by: https://github.com/zou3519
This fixes an issue with mobile: The output of view_copy ops should always be contiguous.
Later, we can consider adding optional arguments to the `view_copy()` functions to let you explicitly say what the contiguity of the output can be (e.g. channels_last)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/85747
Approved by: https://github.com/ezyang
1. Added references `_refs.broadcast_shapes`
2. Added OpInfo test for `torch.broadcast_shapes`
A few minor changes:
- `test_python_ref_meta` and `_ref_test_helper` update to avoid non-tensor outputs
- type annotation update for `_resize_meta`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78612
Approved by: https://github.com/mruberry
We don't have any coverage for meta tensor correctness for backwards
because torch function mode can only allow us to interpose on
Python torch API calls, but backwards invocations happen from C++.
To make this possible, I add torch_dispatch_meta test which runs the
tests with __torch_dispatch__
While doing this, I needed to generate fresh expected failure / skip
lists for the new test suite, and I discovered that my original
scaffolding for this purpose was woefully insufficient. So I rewrote
how the test framework worked, and at the same time rewrote the
__torch_function__ code to also use the new logic. Here's whats
new:
- Expected failure / skip is now done on a per function call basis,
rather than the entire test. This means that separate OpInfo
samples for a function don't affect each other.
- There are now only two lists: expect failure list (where the test
consistently fails on all runs) and skip list (where the test
sometimes passes and fails.
- We explicitly notate the dtype that failed. I considered detecting
when something failed on all dtypes, but this was complicated and
listing everything out seemed to be nice and simple. To keep the
dtypes short, I introduce a shorthand notation for dtypes.
- Conversion to meta tensors is factored into its own class
MetaConverter
- To regenerate the expected failure / skip lists, just run with
PYTORCH_COLLECT_EXPECT and filter on a specific test type
(test_meta or test_dispatch_meta) for whichever you want to update.
Other misc fixes:
- Fix max_pool1d to work with BFloat16 in all circumstances, by making
it dispatch and then fixing a minor compile error (constexpr doesn't
work with BFloat16)
- Add resolve_name for turning random torch API functions into string
names
- Add push classmethod to the Mode classes, so that you can more easily
push a mode onto the mode stack
- Add some more skips for missing LAPACK
- Added an API to let you query if there's already a registration for
a function, added a test to check that we register_meta for all
decompositions (except detach, that decomp is wrong lol), and then
update all the necessary sites to make the test pass.
Signed-off-by: Edward Z. Yang <ezyangfb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/77477
Approved by: https://github.com/zou3519
Fixes#73184
In the issue the output tensor's shape is `[2, 4, 536870912, 536870912]` which results in a `numel()` slightly below the point of overflow. When the storage is created it does `numel() * 8` which overflows and a much smaller storage is allocated than required.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/73719
Approved by: https://github.com/ezyang, https://github.com/malfet
Hi!
The PR aims to fix#70657. The objective was to ensure that `torch.ravel()` returns contiguous outputs for non-contiguous inputs. It also adds the test verifying the contiguity of the `torch.ravel`, which was missing.
I am looking forward to your viewpoints. Thanks :)
Thank you so much, @kshitij12345, for helping me clear up the concepts! :)
cc: @mruberry @kshitij12345
Pull Request resolved: https://github.com/pytorch/pytorch/pull/71771
Approved by: https://github.com/mruberry
Summary:
Partially fixes https://github.com/pytorch/pytorch/issues/66066
This PR:
- cleans up op-specific testing from test_autograd. test_autograd should be reserved for testing generic autograd functionality
- tests related to an operator are better colocated
- see the tracker for details
What to think about when moving tests to their correct test suite:
- naming, make sure its not too generic
- how the test is parametrized, sometimes we need to add/remove a device/dtype parameter
- can this be merged with existing tests
Pull Request resolved: https://github.com/pytorch/pytorch/pull/67413
Reviewed By: jbschlosser, albanD
Differential Revision: D32031480
Pulled By: soulitzer
fbshipit-source-id: 8e13da1e58a38d5cecbfdfd4fe2b4fe6f816897f
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/63602
This PR fixes the case when a read and write is performed on a memory shared between mutable and (or) non-mutable arguments. Example:
```
a=torch.tensor([1+1j])
b=a.conj()
b.add_(a) # should return tensor([2]) but returns tensor ([2-2j])
```
The issue here is that in the conjugate fallback, we resolve the conjugation in-place for mutable arguments which can be a problem as shown above in the case when other input arguments share memory with the mutable argument(s).
This PR fixes this issue by:
1. first scanning through the operator input arguments and creating a vector of mutable arguments that have the conj bit set to `True` (and accordingly setting the flag `check_for_alias_with_mut_arg ` to `True` or `False`).
2. Iterating through all the arguments. At this time we only look at the non-mutable arguments. If `check_for_alias_with_mut_arg` is set to `True`, then we iterate through `mutable_inputs` to check if the current arg tensor in question doesn't alias any of the entries in `mutable_inputs`. If yes, then we clone the non-mutable tensor arg, else we resolve the conjugation as before.
3. Now we look through the mutable_inputs vector (which contains only mutable input tensors with conj bit set to `True`). We in-place conjugate each of the entries in the vector.
4. Do the computation.
5. Re-conjugate the mutable argument tensors.
NOTE: `TensorLists` are not fully handled in ConjugateFallback. Please see the in-line comment for more details.
Fixes https://github.com/pytorch/pytorch/issues/59943
Test Plan: Imported from OSS
Reviewed By: gmagogsfm
Differential Revision: D30466905
Pulled By: anjali411
fbshipit-source-id: 58058e5e6481da04a12d03f743c1491942a6cc9b
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/63554
Following https://github.com/pytorch/pytorch/pull/61840#issuecomment-884087809, this deprecates all the dtype getters publicly exposed in the `torch.testing` namespace. The reason for this twofold:
1. If someone is not familiar with the C++ dispatch macros PyTorch uses, the names are misleading. For example `torch.testing.floating_types()` will only give you `float32` and `float64` skipping `float16` and `bfloat16`.
2. The dtype getters provide very minimal functionality that can be easily emulated by downstream libraries.
We thought about [providing an replacement](https://gist.github.com/pmeier/3dfd2e105842ad0de4505068a1a0270a), but ultimately decided against it. The major problem is BC: by keeping it, either the namespace is getting messy again after a new dtype is added or we need to somehow version the return values of the getters.
Test Plan: Imported from OSS
Reviewed By: H-Huang
Differential Revision: D30662206
Pulled By: mruberry
fbshipit-source-id: a2bdb10ab02ae665df1b5b76e8afa9af043bbf56
