32e052e468
[docs] improve torch.stack example code to be reproducible ( #133857 )
...
Improve the sample code can produce the expected results after copying and executing it.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133857
Approved by: https://github.com/soulitzer
2024-08-21 14:07:02 +00:00
441c1c03d5
Prevent an unnecessary device -> host copy for CuPy arrays when not explicitly setting a device in torch.as_tensor. ( #132595 )
...
See title. Until now, calling `torch.as_tensor` on a CuPy array would return a CPU tensor, when not providing a device. This is most likely not desired.
Fixes #132553
```python3
import torch
import cupy as cp
cupy_arr = cp.asarray([1, 2, 3])
# Default case
t = torch.as_tensor(cupy_arr)
# New behavior, same device as cupy_arr now, was cpu before
print(t.device) # cuda:0
# Explicitly set device
t = torch.as_tensor(cupy_arr, device='cpu')
print(t.device) # cpu
# Implicit default device
torch.set_default_device('cpu')
t = torch.as_tensor(cupy_arr)
print(t.device) # cpu
# Default device via context manager
torch.set_default_device('cuda')
with torch.device('cpu'):
t = torch.as_tensor(cupy_arr)
print(t.device) # cpu
# Unset default device
torch.set_default_device(None)
t = torch.as_tensor(cupy_arr)
# New behavior, same device as cupy_arr now, was cpu before
print(t.device) # cuda:0
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/132595
Approved by: https://github.com/ezyang
2024-08-08 00:26:58 +00:00
4226ed1585
[BE] Format uncategorized Python files with ruff format ( #132576 )
...
Remove patterns `**`, `test/**`, and `torch/**` in `tools/linter/adapters/pyfmt_linter.py` and run `lintrunner`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/132576
Approved by: https://github.com/ezyang , https://github.com/Skylion007
ghstack dependencies: #132574
2024-08-04 17:13:31 +00:00
a4ea776881
Add pinned memory support to sparse COO/CSR/CSC/BSR/BSC tensors ( #129645 )
...
As in the title:
To register indices/values of a sparse XYZ tensor with CUDA, the following methods are supported
- `sparse_xyz_tensor(indices, values, pin_memory=True)`
- `sparse_xyz_tensor(indices, values).pin_memory()`
- `sparse_xyz_tensor(indices.pin_memory(), values.pin_memory())`
Fixes https://github.com/pytorch/pytorch/issues/115330
Pull Request resolved: https://github.com/pytorch/pytorch/pull/129645
Approved by: https://github.com/amjames , https://github.com/cpuhrsch , https://github.com/eqy
2024-08-02 08:55:55 +00:00
9f6d7df3d9
docs(multinomial): Add reference to Multinomial class ( #131904 )
...
This PR just adds the reference to the class
`torch.distributions.multinomial.Multinomial` in `torch.multinomial`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/131904
Approved by: https://github.com/jbschlosser
2024-07-30 18:05:07 +00:00
9cae2160f5
Introduce the concept of Accelerators to PyTorch doc ( #129363 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/129363
Approved by: https://github.com/EikanWang , https://github.com/gujinghui , https://github.com/albanD
ghstack dependencies: #129463 , #129205
2024-07-15 14:24:46 +00:00
634b62f111
typing proxy_tensor.py ( #129182 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/129182
Approved by: https://github.com/Chillee
2024-07-12 23:17:09 +00:00
1d93367cfa
Fix typo ( #130305 )
...
Fixes #130241
that is a reopen pr of #130244 , for possibly fixing the failed job
Pull Request resolved: https://github.com/pytorch/pytorch/pull/130305
Approved by: https://github.com/Skylion007
2024-07-09 20:02:00 +00:00
735044191f
[Easy] Add whitespace after comma when re-rendering tuple default value in schema ( #129884 )
...
The default value of `rot90()` in the schema registry is `[0,1]` because we split the function schema by `", "`. There should be no space after `,` in `[0,1]`.
5c9d5272e4/aten/src/ATen/native/native_functions.yaml (L6120-L6126)https://github.com/pytorch/pytorch/pull/129884
Approved by: https://github.com/ezyang
2024-07-03 11:45:24 +00:00
75f64e1203
Fix test test_type_hints.py::TestTypeHints::test_doc_examples ( #129829 )
...
As per the title, this test was broken for months.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/129829
Approved by: https://github.com/ezyang
2024-07-01 13:28:37 +00:00
a70a7337d2
Update torch.nanmean() docstring to mention input dtype requirement ( #128155 )
...
Fixes #120570
## Description
Update torch.nanmean() docstring to mention input dtype requirement as either floating point type or complex.
Previously, the torch.mean() docstring had been updated in #120208 in a similar manner, but the torch.nanmean() docstring was not updated.
## Checklist
- [X] The issue that is being fixed is referred in the description.
- [X] Only one issue is addressed in this pull request.
- [x] Labels from the issue that this PR is fixing are added to this pull request.
- [X] No unnecessary issues are included into this pull request.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128155
Approved by: https://github.com/malfet
2024-06-12 17:46:36 +00:00
afe15d2d2f
Flip default value for mypy disallow_untyped_defs [3/11] ( #127840 )
...
See #127836 for details.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/127840
Approved by: https://github.com/oulgen
2024-06-08 18:28:01 +00:00
3591bce6c7
Add usage explanation in torch.dot ducment ( #125908 )
...
Fixes #125842
Add unsupported declaration on <code>torch.dot</code>, avoid misused like:
```python
>>> t1, t2 = torch.tensor([0,1]), torch.tensor([2,3])
>>> torch.dot(input=t1, other=t2)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: dot() missing 1 required positional arguments: "tensor"
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125908
Approved by: https://github.com/albanD
2024-05-22 20:33:12 +00:00
4f1a56cd42
Switched from parameter in can_cast to from_. ( #126030 )
...
Fixes #126012 .
`from` is a reserved keyword in Python, thus we can't make the C++ impl available with `from` as function parameter. This PR changes the name to `from_` and also adjusts the docs.
If we want to preserve backwards compatibility, we can leave the C++ name as-is and only fix the docs. However, `torch.can_cast(from_=torch.int, to=torch.int)` won't work then.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126030
Approved by: https://github.com/albanD
2024-05-16 20:58:24 +00:00
6065a4d46e
Revert "Switched from parameter in can_cast to from_. ( #126030 )"
...
This reverts commit 06d6bb4ebabc64433224970024ada1781508197d.
Reverted https://github.com/pytorch/pytorch/pull/126030 on behalf of https://github.com/huydhn due to Sorry for reverting your change but i need to revert it to avoid a diff train conflict with https://github.com/pytorch/pytorch/pull/125995 . Please help rebase and I will reland the change ([comment](https://github.com/pytorch/pytorch/pull/126030#issuecomment-2113757469 ))
2024-05-16 01:42:23 +00:00
06d6bb4eba
Switched from parameter in can_cast to from_. ( #126030 )
...
Fixes #126012 .
`from` is a reserved keyword in Python, thus we can't make the C++ impl available with `from` as function parameter. This PR changes the name to `from_` and also adjusts the docs.
If we want to preserve backwards compatibility, we can leave the C++ name as-is and only fix the docs. However, `torch.can_cast(from_=torch.int, to=torch.int)` won't work then.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126030
Approved by: https://github.com/albanD
2024-05-16 00:09:54 +00:00
4d8fa7df40
Fix four misspellings of "its" in documentation ( #125681 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125681
Approved by: https://github.com/Skylion007 , https://github.com/svekars
2024-05-13 18:14:09 +00:00
c0c2f6156a
Updated docs to add the error case for torch.multinomial Issue#125388 ( #125495 )
...
Summary: Updated docs to add the error condition for torch.multinomial
Test Plan: No change in code
Reviewers: @drisspg
Subscribers: @drisspg
Tasks:
Tags:
Fixes #125388
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125495
Approved by: https://github.com/drisspg
2024-05-07 00:26:27 +00:00
a8574a9719
Fix global flake8 issues ( #124771 )
...
Prior to this `lintrunner --all-files --take FLAKE8` failed.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/124771
Approved by: https://github.com/Skylion007
ghstack dependencies: #124428
2024-04-26 15:35:53 +00:00
1ac60484c1
Revert "Fix global flake8 issues ( #124771 )"
...
This reverts commit f01275934bfa1ff358b1c01d3754f2807cd04ee2.
Reverted https://github.com/pytorch/pytorch/pull/124771 on behalf of https://github.com/jeanschmidt due to Unfortunately, I needed to revert #123735 and this one depends on it. So please check if there are no merge conflicts or breakages and feel free to merge this PR again ([comment](https://github.com/pytorch/pytorch/pull/124428#issuecomment-2078699836 ))
2024-04-26 06:15:17 +00:00
f01275934b
Fix global flake8 issues ( #124771 )
...
Prior to this `lintrunner --all-files --take FLAKE8` failed.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/124771
Approved by: https://github.com/Skylion007
ghstack dependencies: #124428
2024-04-25 14:25:00 +00:00
249e65b92d
Graph-Safe RNG State Exchange for Tensor Parallelism ( #114068 )
...
See #113541
The PR allows for registering and controlling multiple RNG states using indices, ensuring cudagraph-safe operations, and includes both C++ and Python API changes to support this functionality.
cc @eellison @anijain2305 @jansel @ezyang @ptrblck @csarofeen @mcarilli
Pull Request resolved: https://github.com/pytorch/pytorch/pull/114068
Approved by: https://github.com/ezyang , https://github.com/eqy , https://github.com/xuzhao9
2024-03-27 01:14:38 +00:00
4dc09d6aa4
Revert "Graph-Safe RNG State Exchange for Tensor Parallelism ( #114068 )"
...
This reverts commit e9dcda5cba92884be6432cf65a777b8ed708e3d6.
Reverted https://github.com/pytorch/pytorch/pull/114068 on behalf of https://github.com/ezyang due to memory leak in another ci ([comment](https://github.com/pytorch/pytorch/pull/114068#issuecomment-2018044527 ))
2024-03-25 13:49:04 +00:00
e9dcda5cba
Graph-Safe RNG State Exchange for Tensor Parallelism ( #114068 )
...
See #113541
The PR allows for registering and controlling multiple RNG states using indices, ensuring cudagraph-safe operations, and includes both C++ and Python API changes to support this functionality.
cc @eellison @anijain2305 @jansel @ezyang @ptrblck @csarofeen @mcarilli
Pull Request resolved: https://github.com/pytorch/pytorch/pull/114068
Approved by: https://github.com/ezyang
2024-03-21 01:57:08 +00:00
680cfec295
Fix the default value of side in torch.searchsorted ( #120066 )
...
Fixes #119999 , currently the [doc](https://pytorch.org/docs/stable/generated/torch.searchsorted.html#torch.searchsorted ) shows the default value of `side = "left"`
<img width="600" alt="Screenshot 2024-02-16 at 10 36 08 AM" src="https://github.com/pytorch/pytorch/assets/7495155/e7d159aa-4985-4f50-9d81-6e71c3116c0d ">
while the [implementation ](https://github.com/pytorch/pytorch/blob/main/aten/src/ATen/native/native_functions.yaml#L11247 ) gives the default value of `side = c10::nullopt`.
- fix the [torch doc](https://github.com/pytorch/pytorch/blob/main/torch/_torch_docs.py#L13782 ) such that the default value of side is None.
- fix the [comment in cpp](4dc75f9084/aten/src/ATen/native/Bucketization.cpp (L19)https://github.com/pytorch/pytorch/pull/120066
Approved by: https://github.com/malfet
2024-02-22 19:35:17 +00:00
bb72bfe2ac
Add code example for torch.stack() ( #120304 )
...
Fixes #120303
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120304
Approved by: https://github.com/albanD
2024-02-22 18:30:30 +00:00
d336be2942
Update torch.mean() description about dtype restriction. ( #120208 )
...
Fixes #120173
Co-authored-by: Jeffrey Wan <soulitzer@gmail.com >
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120208
Approved by: https://github.com/soulitzer
2024-02-21 18:04:11 +00:00
2d4aa91a10
Fix searchsorted function signature in docs ( #120086 )
...
Side should be optional string, to match definition in native_functions: fbe8e0f92d/aten/src/ATen/native/native_functions.yaml (L11246)https://github.com/pytorch/pytorch/issues/119999
Test plan: https://docs-preview.pytorch.org/pytorch/pytorch/120086/generated/torch.searchsorted.html#torch-searchsorted
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120086
Approved by: https://github.com/lezcano
2024-02-16 20:00:04 +00:00
0fd371c868
fix torch.cumsum docs ( #117944 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/117944
Approved by: https://github.com/zou3519
2024-02-13 15:29:06 +00:00
28c30f29be
Update documentation for set_flush_denormal support on ARM ( #119354 )
...
**Documentation update for set_flush_denormal():**
-> set_flush_denormal() is now supported on ARM CPU's.
-> **PR:** https://github.com/pytorch/pytorch/pull/115184 (Already merged)
**Reference page:** https://pytorch.org/docs/stable/generated/torch.set_flush_denormal.html
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119354
Approved by: https://github.com/drisspg
2024-02-12 20:53:22 +00:00
31f00b0160
Clarify that legacy cat behavior only applies for 1-D tensor ( #119684 )
...
Signed-off-by: Edward Z. Yang <ezyang@meta.com >
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119684
Approved by: https://github.com/albanD
2024-02-12 18:13:04 +00:00
5f2ad407a9
Fix typo on torch.frombuffer() documentation ( #119214 )
...
Fixes #114345
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119214
Approved by: https://github.com/albanD
2024-02-07 00:41:51 +00:00
7bbd9befed
Improve example for `torch.mode()` ( #115308 )
...
Fixes #89820 and improves the documentation.
Co-authored-by: Sam Gross <colesbury@gmail.com >
Pull Request resolved: https://github.com/pytorch/pytorch/pull/115308
Approved by: https://github.com/colesbury
2024-02-03 00:13:26 +00:00
8ff55c7e68
Clarified sampling process of torch.randn for complex dtypes. ( #118315 )
...
Fixes #118269 .
Clarified the docs of `torch.randn` and `torch.randn_like`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118315
Approved by: https://github.com/lezcano
2024-01-26 13:05:19 +00:00
75818adcf7
Pyi doc inclusion + fix ( #117267 )
...
Reland of https://github.com/pytorch/pytorch/pull/114705 with extra fix to smoothly handle when the modules we're trying to load are not available (and thus the pyi won't contain the docs in this case).
Tested locally that it works properly in fbcode.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/117267
Approved by: https://github.com/ezyang
2024-01-15 13:06:53 +00:00
767e1b6349
Revert "Bring docstring to .pyi file ( #114705 )"
...
This reverts commit 0dd5deecedd136852c7ccc81630eaefbebe5be29.
Reverted https://github.com/pytorch/pytorch/pull/114705 on behalf of https://github.com/facebook-github-bot due to Diff reverted internally ([comment](https://github.com/pytorch/pytorch/pull/114705#issuecomment-1887165326 ))
2024-01-11 13:30:44 +00:00
0dd5deeced
Bring docstring to .pyi file ( #114705 )
...
Fixes #37762
Since the original issue hasn't been making progress for more than 3 years, I am attempting to make this PR to at least make some progress forward.
This PR attempts to add docstring to the `.pyi` files. The docstrings are read from [`_torch_docs`](https://github.com/pytorch/pytorch/blob/main/torch/_torch_docs.py ) by mocking [`_add_docstr`](9f073ae304/torch/csrc/Module.cpp (L329)https://github.com/pytorch/pytorch/files/13494263/_VariableFunctions.pyi.txt )
<img width="787" alt="image" src="https://github.com/pytorch/pytorch/assets/6421097/73c2e884-f06b-4529-8301-0ca0b9de173c ">
And the docstring can be picked up by VSCode:
<img width="839" alt="image" src="https://github.com/pytorch/pytorch/assets/6421097/1999dc89-a591-4c7a-80ac-aa3456672af4 ">
<img width="908" alt="image" src="https://github.com/pytorch/pytorch/assets/6421097/ecf3fa92-9822-4a3d-9263-d224d87ac288 ">
Pull Request resolved: https://github.com/pytorch/pytorch/pull/114705
Approved by: https://github.com/albanD
2024-01-09 18:37:16 +00:00
43f42bf3cb
Updated docs for deprecated torch.set_default_tensor_type ( #115041 )
...
Added deprecation note for torch.set_default_tensor_type. Updated docs that referenced this method.
Fixes #113646 .
Pull Request resolved: https://github.com/pytorch/pytorch/pull/115041
Approved by: https://github.com/janeyx99
2023-12-07 16:17:36 +00:00
c678c5ef38
[doc] caution torch.multinomial usage ( #112892 )
...
Fixes #107406
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112892
Approved by: https://github.com/albanD
2023-11-15 18:20:48 +00:00
fd209543d5
Add torch.utils.deterministic.fill_uninitialized_memory flag ( #111377 )
...
Part of #109802
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111377
Approved by: https://github.com/albanD , https://github.com/aaronenyeshi
2023-11-01 16:10:09 +00:00
ace2713d1e
Revert "Add torch.utils.deterministic.fill_uninitialized_memory flag ( #111377 )"
...
This reverts commit f1785373c08b9e8383b7eec3391d57053209b525.
Reverted https://github.com/pytorch/pytorch/pull/111377 on behalf of https://github.com/facebook-github-bot due to Diff reverted internally ([comment](https://github.com/pytorch/pytorch/pull/111377#issuecomment-1784179040 ))
2023-10-29 17:41:55 +00:00
f1785373c0
Add torch.utils.deterministic.fill_uninitialized_memory flag ( #111377 )
...
Part of #109802
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111377
Approved by: https://github.com/albanD
2023-10-26 02:39:06 +00:00
b54ab57522
Document torch.from_file and fix UntypedStorage.from_file docs ( #111688 )
...
Fixes https://github.com/pytorch/pytorch/issues/37439
Also threads through filename so it is accessible via `t.storage().filename`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111688
Approved by: https://github.com/albanD
2023-10-25 19:28:11 +00:00
46e80ce58a
[ATen] Support multi dim any and all reductions ( #110310 )
...
This adds a new overload to `all` and `any` with support for multiple reduction dims.
```
all.dims(Tensor self, int[1]? dim=None, bool keepdim=False) -> Tensor
any.dims(Tensor self, int[1]? dim=None, bool keepdim=False) -> Tensor
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/110310
Approved by: https://github.com/lezcano , https://github.com/albanD , https://github.com/justinchuby
2023-10-24 21:33:53 +00:00
93e5065ba0
[CODEMOD][caffe2] replace numpy.bool with bool ( #111432 )
...
Test Plan:
numpy.bool is long deprecated and removed starting numpy-1.20.0 [1]. This replaces all references with equivalent `bool` type using the following oneliner:
```
rg -l 'np\.bool' caffe2 | grep '\.py$' | xargs perl -pi -e 's,\bnp\.bool\b,bool,'
```
1. https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
Differential Revision: D50372711
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111432
Approved by: https://github.com/Skylion007
2023-10-18 18:56:40 +00:00
d8de45d22c
Update arg{min,max} tests and docs ( #110845 )
...
The `argmin` docs had been updated in
https://github.com/pytorch/pytorch/issues/78791 but left a minor typo.
`argmax` had a similar issue but was not noticed.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/110845
Approved by: https://github.com/eellison
2023-10-13 21:40:29 +00:00
9f0601df6d
Fix a typo in cholesky_inverse documentation ( #110364 )
...
Very small PR to fix a typo in [https://pytorch.org/docs/stable/generated/torch.cholesky_inverse.html ](cholesky_inverse) doc.
According to the current doc, the function expects $A$, the symmetric positive-definite matrix, as input. But the examples given (and more important, the code) is using $u$ the cholesky decomposition of this matrix (like cholesky_solve).
Also, it provides a correct example of batch usage of this function.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/110364
Approved by: https://github.com/lezcano
2023-10-04 12:30:11 +00:00
7b53303d3c
Improved the docs for torch.std, torch.var, torch.std_mean, torch.var_mean and torch.cov ( #109326 )
...
Fixes #109186 .
This PR updates the docs for
- `torch.var`
- `torch.var_mean`
- `torch.std`
- `torch.std_mean`
- `torch.cov`
to reflect the actual implementation behavior when `correction >= N`. The math for `torch.cov` should probably be double checked before merging.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109326
Approved by: https://github.com/albanD
2023-09-19 20:47:24 +00:00
2f53bca0fc
[Docs] Fix typo in torch.unflatten ( #109588 )
...
Fixes https://github.com/pytorch/pytorch/issues/109559
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109588
Approved by: https://github.com/lezcano
2023-09-19 10:37:45 +00:00
a6d34c60a1
Fixing searchsorted doc ( #109364 )
...
Removing ambiguous description
Fixes #109298
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109364
Approved by: https://github.com/colesbury
2023-09-18 23:12:53 +00:00
