OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 00:21:07 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity
47,378 Commits 8,304 Branches 150 Tags 1.5 GiB
a732bbea23
Commit Graph

15125 Commits

Author SHA1 Message Date
kshitij12345
a732bbea23 [meta] Add meta support for fft ops (#79311) 
As per title
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79311
Approved by: https://github.com/ezyang
 2022-06-13 01:56:42 +00:00
kshitij12345
bd1a35dfc8 [meta] diag ops, trace (#79341) 
meta registration for `diag.out` and update test skips/expectedFailures
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79341
Approved by: https://github.com/ezyang
 2022-06-12 18:45:03 +00:00
Edward Z. Yang
213a8fc992 Put symint overloads on a different name 
Due to implicit conversion shenanigans, having both IntArrayRef
and SymIntArrayRef overloads makes {} ambiguous.  While we could
fix this by making a single unified type that accepts all the overloads
we want, an easier fix was to just push the SymIntArrayRef overload
to its own name.

Signed-off-by: Edward Z. Yang <ezyangfb.com>

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79281

Approved by: https://github.com/suo
 2022-06-12 14:36:39 +00:00
Taylor Robie
c321dfe1b5 move tree tests to the start of test_profiler.py 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79301

Approved by: https://github.com/davidchencsl
 2022-06-12 04:24:30 +00:00
pritam
a81be44410 Fix shard_module to appropriately deal with sub process groups. 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79264

`shard_module` API didn't work correctly with a sub-pg since
`dist.scatter` actually takes the global rank as input for `src`.

Fixing this by passing in the appropriate rank to `dist.scatter`

Differential Revision: [D37062766](https://our.internmc.facebook.com/intern/diff/D37062766/)

Approved by: https://github.com/fduwjj, https://github.com/wanchaol
 2022-06-12 03:50:45 +00:00
PyTorch MergeBot
5413580f9e Revert "[JIT] Propagate profiled information to DifferentiableGraph outputs" 
This reverts commit 1d2a6c2e94.

Reverted https://github.com/pytorch/pytorch/pull/78875 on behalf of https://github.com/davidberard98 due to Internal failures were bisected to this change
 2022-06-12 00:14:08 +00:00
Michael Suo
30fb2c4aba [lint] autoformat test/cpp and torch/csrc 
Let's have some fun.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/78828

Approved by: https://github.com/ezyang
 2022-06-11 21:11:16 +00:00
Mikayla Gawarecki
1ec30a6647 Add offsets-based reduction to segment_reduce (CPU, CUDA) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78907

Approved by: https://github.com/cpuhrsch
 2022-06-11 17:43:42 +00:00
Michael Suo
c978b609f7 [ci] remove IN_CI env var 
The conventional env var to set is CI. Both circle and GHA set it, so
IN_CI is unnecessary

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79229

Approved by: https://github.com/janeyx99
 2022-06-11 17:16:30 +00:00
lezcano
9fc2518a8a Update and improve the heuristics for linalg.lu_solve 
This PR adds getrf_cublas to the functions considered in the heuristics
for lu_solve.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/73878

Approved by: https://github.com/nikitaved, https://github.com/IvanYashchuk, https://github.com/mruberry
 2022-06-11 04:06:40 +00:00
lezcano
54949a5abc Simplify and optimize linalg.solve 
This PR heavily simplifies the code of `linalg.solve`. At the same time,
this implementation saves quite a few copies of the input data in some
cases (e.g. A is contiguous)

We also implement it in such a way that the derivative goes from
computing two LU decompositions and two LU solves to no LU
decompositions and one LU solves. It also avoids a number of unnecessary
copies the derivative was unnecessarily performing (at least the copy of
two matrices).

On top of this, we add a `left` kw-only arg that allows the user to
solve `XA = B` rather concisely.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/74046

Approved by: https://github.com/nikitaved, https://github.com/IvanYashchuk, https://github.com/mruberry
 2022-06-11 04:06:40 +00:00
Akshay Parashar
65a37923f9 [Static Runtime] Exception handling during fork subgraph execution (#79292) 
Summary:
- Exception handling was not performed in forked subgraph execution
- forked subgraph runtime can throw runtime exception. Future returned by prim::fork needs to handle exceptions so that aten::wait handles it.

Test Plan:
local test cases:
  - buck test caffe2/benchmarks/static_runtime/fb:test_fb_operators
  - buck test mode/opt caffe2/benchmarks/static_runtime:static_runtime_cpptest
  - buck test mode/opt caffe2/test:static_runtime

Async execution of the subgraph is tested by adding pytorch profiler hooks on the StaticRuntime execution via below code. Async execution in threadpool is verfiied by checking trace
  with profile(activities=[ProfilerActivity.CPU]) as prof:
      static_runtime_module(inputs)
  prof.export_chrome_trace("trace.json")

Differential Revision: D37072493

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79292
Approved by: https://github.com/mikeiovine
 2022-06-11 03:11:49 +00:00
Michael Andreas Dagitses
ab2ca95dd1 turn on -Werror=unused-variable in our Bazel CPU build 
Summary:
We also fix any existing issues. Note that we only do this for the CPU
build because nvcc is considered a C++ toolchain but it does not have
the same flag support. Adding flags to the GPU build will cause nvcc
errors.

Test Plan: Built locally, rely on CI to confirm.

Reviewers: malfet

Subscribers:

Tasks:

Tags:

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79156

Approved by: https://github.com/seemethere, https://github.com/osalpekar, https://github.com/albanD
 2022-06-11 02:46:34 +00:00
Mikayla Gawarecki
e727539c29 Support multi-dimensional lengths in segment_reduce to support pytorch_scatter.segment_* functionalities (CUDA) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/77061

Approved by: https://github.com/cpuhrsch
 2022-06-11 01:45:22 +00:00
anjali411
38350acf8f Autogen Tags enum, and allow specifying tags while defining an op 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79322

Approved by: https://github.com/albanD
 2022-06-11 00:29:32 +00:00
drisspg
bdcee8f995 update is_same_size to work with nested tensor dispatch 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79297

Approved by: https://github.com/soulitzer
 2022-06-11 00:07:27 +00:00
mingfeima
50e1301293 qcat: use direct memcpy when all the inputs and output share the same scale and zero_point 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/71903

Approved by: https://github.com/jerryzh168
 2022-06-10 23:59:14 +00:00
Michael Andreas Dagitses
606b234336 turn on -Werror=unused-function in our Bazel CPU build 
Summary:
We also fix any existing issues. Note that we only do this for the CPU
build because nvcc is considered a C++ toolchain but it does not have
the same flag support. Adding flags to the GPU build will cause nvcc
errors.

Test Plan: Built locally, rely on CI to confirm.

Reviewers: malfet

Subscribers:

Tasks:

Tags:

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79154

Approved by: https://github.com/seemethere, https://github.com/osalpekar, https://github.com/albanD
 2022-06-10 22:11:54 +00:00
PyTorch MergeBot
d28e9e145b Revert "[nvfuser_upstream_push] nvfuser code base bump 060822 (#79147)" 
This reverts commit 49c41b87a2.

Reverted https://github.com/pytorch/pytorch/pull/79147 on behalf of https://github.com/janeyx99 due to Broke 11.3 builds on trunk 49c41b87a2
 2022-06-10 20:55:10 +00:00
PyTorch MergeBot
bcd7a20953 Revert "turn on -Werror=unused-function in our Bazel CPU build" 
This reverts commit 67d313a032.

Reverted https://github.com/pytorch/pytorch/pull/79154 on behalf of https://github.com/malfet due to Breaks bazel build: 67d313a032
 2022-06-10 20:43:03 +00:00
PyTorch MergeBot
b712467cd1 Revert "Add mutation checks for tensor inputs" 
This reverts commit 83c0a2bc38.

Reverted https://github.com/pytorch/pytorch/pull/79078 on behalf of https://github.com/davidberard98 due to broke bazel build-and-test, see [https://github.com/pytorch/pytorch/runs/6836001002?check_suite_focus=true](https://github.com/pytorch/pytorch/runs/6836001002?check_suite_focus=true%22)
 2022-06-10 20:15:30 +00:00
kshitij12345
7b307e5fca [meta] angle, angle.out (#79278) 
meta registration for `angle, angle.out`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79278
Approved by: https://github.com/anjali411
 2022-06-10 20:06:31 +00:00
jjsjann123
49c41b87a2 [nvfuser_upstream_push] nvfuser code base bump 060822 (#79147) 
Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/

Bug fixes and minor refactor

Squashed commits to WAR github API
Commits that's actually in this PR from the devel branch:

```
4c60e7dff22a494632370e5df55c011007340d06 Add examples infrastructure for using nvFuser in a standalone program (#1725)
02a05d98334ffa580d73ccb28fdb8c577ad296fe Fix issue #1751 (#1753)
8a69aa320bd7629e1709fe5ceb7104d2c88ec84c Refactor NvFuser transpose API to match eager mode behavior (#1746)
ffdf6b7709048170d768217fcd7083fc8387f932 Remove BroadcastWithoutStride. (#1738)
02bab16035e70734450c02124f5cdaa95cf5749d Fix flipping of a boolean flag (#1745)
465d66890c8242e811224359cbdb1c2915490741 cleanup (#1744)
26d354e68720bc7dd2d3b1338ac01b707a230b6a fixing noncontig broadcast (#1742)
856b6b2f9073662dd98ca22ba6c3540e20eb1cdd Add IterDomainBuilder (#1736)
1fd974f912cd4c1e21cbd16e2abb23598d66a02f fixing warning for gcc7 (#1732)
de2740a43a869f8272c2648e091d7b8235097db9 disabling complex in python tests for #1730 (#1733)
fbbbe0a2e7c7a63e0e2719b8bfccb759b714221a fixing MSVC build (#1728)
b5feee5e2b28be688dbddc766f3c0220389c8175 Fix the fused reduction runtime kernel (#1729)
5247682dff5980bb66edf8d3aac25dea2ef2ced5 Re-entrant GroupedGridReduction (#1727)
```

RUN_TORCHBENCH: nvfuser
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79147
Approved by: https://github.com/davidberard98
 2022-06-10 19:37:42 +00:00
samdow
5e926aafab add utils for checking that all modes are in the same scope and finding the outermost mode 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78847

Approved by: https://github.com/ezyang, https://github.com/zou3519
 2022-06-10 19:31:05 +00:00
Rohan Varma
ec86070922 Checkpoint util 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78704

Approved by: https://github.com/zhaojuanmao
 2022-06-10 18:37:36 +00:00
Michael Andreas Dagitses
67d313a032 turn on -Werror=unused-function in our Bazel CPU build 
Summary:
We also fix any existing issues. Note that we only do this for the CPU
build because nvcc is considered a C++ toolchain but it does not have
the same flag support. Adding flags to the GPU build will cause nvcc
errors.

Test Plan: Built locally, rely on CI to confirm.

Reviewers: malfet

Subscribers:

Tasks:

Tags:

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79154

Approved by: https://github.com/seemethere, https://github.com/osalpekar, https://github.com/albanD
 2022-06-10 18:30:08 +00:00
samdow
3734fcc8f8 add ability to push a mode if the current mode is an ancestor 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78822

Approved by: https://github.com/ezyang, https://github.com/zou3519
 2022-06-10 18:27:04 +00:00
goldenxuett
83c0a2bc38 Add mutation checks for tensor inputs 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79078

Approved by: https://github.com/davidberard98, https://github.com/Krovatkin
 2022-06-10 18:17:33 +00:00
Brian Hirsh
7b3a0ff87a Port index.Tensor to structured kernels. 
Tracking issue: #55070

Pull Request resolved: https://github.com/pytorch/pytorch/pull/69607

Approved by: https://github.com/bdhirsh
 2022-06-10 17:27:47 +00:00
George Qi
a90f006fe5 add strides to slow path 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78610

Approved by: https://github.com/ezyang
 2022-06-10 16:59:14 +00:00
Peter Bell
7843a5e882 Move Tensor.grad back into C++ 
`Tensor.grad` was moved to python in #30531 to add a warning. However,
that warning has since been lowered into C++ so this wrapper is no
longer necessary.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/76675

Approved by: https://github.com/albanD
 2022-06-10 13:44:45 +00:00
Kulin Seth
77b6885a22 MPS: add layer_norm_backward (#79189) 
Layernorm backward

Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79189
Approved by: https://github.com/razarmehr, https://github.com/albanD
 2022-06-10 13:25:41 +00:00
Kulin Seth
83239351c5 MPS: add exponential op (#79188) 
Add exponential distribution

Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79188
Approved by: https://github.com/razarmehr, https://github.com/albanD
 2022-06-10 13:16:21 +00:00
PyTorch MergeBot
c99ea0db46 Revert "[PyTorch] Record Sequence Number to Match Forward and Backward Operators (#78795)" 
This reverts commit a299a2fa26.

Reverted https://github.com/pytorch/pytorch/pull/78795 on behalf of https://github.com/janeyx99 due to Broke profiler tests a299a2fa26
 2022-06-10 13:11:44 +00:00
PyTorch MergeBot
d2200e38f7 Revert "fix _unsafe_view schema to work with functionalization" 
This reverts commit 46234df5f1.

Reverted https://github.com/pytorch/pytorch/pull/79148 on behalf of https://github.com/janeyx99 due to Broke 11.3 tests on trunk and on PR, see 46234df5f1
 2022-06-10 13:09:00 +00:00
Michael Andreas Dagitses
f96d96a7fc turn on -Werror=type-limits in our Bazel CPU build 
Summary:
We also fix any existing issues.

Test Plan: Built locally, rely on CI to confirm.

Reviewers: malfet

Subscribers:

Tasks:

Tags:

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79139

Approved by: https://github.com/seemethere, https://github.com/osalpekar, https://github.com/albanD
 2022-06-10 10:04:08 +00:00
vspenubarthi
28c541776c [ao] Added fx model report per_channel detector 
Summary: This code is meant to be a tool to help people get the most out
of their backend by hinting them to use per_channel quantization if it's
supported, which will help increase accuracy significantly. The code is
completed and ready to be reviewed.

Test Plan: test/quantization/fx/test_model_report_fx.py

Reviewers:

Subscribers:

Tasks:

Tags:

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79104

Approved by: https://github.com/HDCharles
 2022-06-10 08:09:59 +00:00
pritam
b9e3d722c4 Use appropriate dtype for sharded linear implementation. 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79255

We use several collective operations in our sharded linear
implementation and for many collectives, we do not set the `dtype` of the
output tensor appropriately. As a result, using a datatype like torch.float16
(which is not the default torch.float32) results in errors.

Fixing this across the board and adding appropriate tests.

Differential Revision: [D37059752](https://our.internmc.facebook.com/intern/diff/D37059752/)

Approved by: https://github.com/fduwjj, https://github.com/wanchaol
 2022-06-10 07:32:15 +00:00
Louis Feng
a299a2fa26 [PyTorch] Record Sequence Number to Match Forward and Backward Operators (#78795) 
Summary: Add sequence number to map forward and backward operators.

Test Plan:
```
buck build mode/dev-nosan cea/ml_perf_model/gpu/scripts: --show-output
buck-out/gen/caffe2/test/profiler#binary.par test_profiler.TestExecutionGraph.test_execution_graph_start_stop
```

Outputs with seq_id: P505545974

Differential Revision: D36881999

Pull Request resolved: https://github.com/pytorch/pytorch/pull/78795
Approved by: https://github.com/robieta
 2022-06-10 05:51:17 +00:00
PyTorch MergeBot
fefff54cad Revert "Revert "Revert "Added {logical_not, trace} refs, moved logical ops to use method overloads""" 
This reverts commit a2d2981e8e.

Reverted https://github.com/pytorch/pytorch/pull/79224 on behalf of https://github.com/suo due to broke lots of things a2d2981e8e
 2022-06-10 04:40:43 +00:00
Horace He
a2d2981e8e Revert "Revert "Added {logical_not, trace} refs, moved logical ops to use method overloads"" 
This reverts commit d67309aefb.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79224

Approved by: https://github.com/mruberry
 2022-06-10 03:07:14 +00:00
PyTorch MergeBot
87a5ecced2 Revert "Support multi-dimensional lengths in segment_reduce to support pytorch_scatter.segment_* functionalities (CUDA)" 
This reverts commit 40f7ef1f3d.

Reverted https://github.com/pytorch/pytorch/pull/77061 on behalf of https://github.com/janeyx99 due to Broke segment_reduce tests on trunk, e.g., 40f7ef1f3d
 2022-06-10 01:57:34 +00:00
Brian Hirsh
46234df5f1 fix _unsafe_view schema to work with functionalization 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79148

Approved by: https://github.com/albanD
 2022-06-10 01:45:04 +00:00
David Berard
1d2a6c2e94 [JIT] Propagate profiled information to DifferentiableGraph outputs 
Without profiled outputs, autodiff can't tell whether or not the outputs of a DifferentiableGraph should requires_grad. Autodiff would default to requires_grad=True if there was no profiled information, causing autodiff to mark tensors as requires_grad when they shouldn't have. This adds requires_grad info onto the type of the output, if it can be found in later uses of the output.

Adds a test for correct autodiff requires_grad behavior and also a test to make sure the output type is correctly annotated in create_autodiff_subgraphs.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/78875

Approved by: https://github.com/eellison
 2022-06-10 00:54:11 +00:00
Mikayla Gawarecki
40f7ef1f3d Support multi-dimensional lengths in segment_reduce to support pytorch_scatter.segment_* functionalities (CUDA) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/77061

Approved by: https://github.com/cpuhrsch
 2022-06-10 00:49:37 +00:00
Elias Ellison
13a8867c01 Add Dynamic Output Shape Tagdfor ata-dependent ops, handle in FakeTensor 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79170

Approved by: https://github.com/ezyang
 2022-06-09 22:16:16 +00:00
Joel Benjamin Schlosser
70d6446a3d Support both train / eval modes for ModuleInfo 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78735

Approved by: https://github.com/albanD
 2022-06-09 20:57:17 +00:00
Animesh Jain
79f18c1aee Minor FX test fix for TorchDynamo (#79206) 
`torch.nn.Identity` gets optimized away with TorchDynamo. Replacing with `torch.nn.tanh`.

@jansel
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79206
Approved by: https://github.com/jansel
 2022-06-09 20:36:04 +00:00
Taylor Robie
84b9e5ba84 Move test_profiler tests to tree rather than icicle format 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/79175

Approved by: https://github.com/ezyang
 2022-06-09 19:45:02 +00:00
Taylor Robie
9f2e2aa28b Revert "Revert "[Profiler] Move python tracing to unified event type (Part 2)"" 
This reverts commit 4305f8e9bd.

replace TEST_CUDA with torch.has_cuda

Pull Request resolved: https://github.com/pytorch/pytorch/pull/79173

Approved by: https://github.com/ezyang
 2022-06-09 19:45:02 +00:00