Commit Graph

29791 Commits

Author SHA1 Message Date
Eli Uriegas d62994a94d ci: Add anaconda pruning to CI pipeline (#44651)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44651

Adds pruning for our anaconda channels (pytorch-nightly, pytorch-test)
into our CI pipeline so that it gets run on a more consistent basis.

Signed-off-by: Eli Uriegas <eliuriegas@fb.com>

Test Plan: Imported from OSS

Reviewed By: walterddr

Differential Revision: D23692851

Pulled By: seemethere

fbshipit-source-id: fa69b506b73805bf2ffbde75d221aef1ee3f753e
2020-09-15 10:51:05 -07:00
kshitij12345 1d733d660d [docs] torch.min/max: remove incorrect warning from docs (#44615)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/44195

cc: mruberry

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44615

Reviewed By: ngimel

Differential Revision: D23703525

Pulled By: mruberry

fbshipit-source-id: 471ebd764be667e29c03a30f3ef341440adc54d2
2020-09-15 10:42:08 -07:00
Xiang Gao 6bc77f4d35 Use amax/maximum instead of max in optimizers (#43797)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/43797

Reviewed By: malfet

Differential Revision: D23406641

Pulled By: mruberry

fbshipit-source-id: 0cd075124aa6533b21375fe2c90c44a5d05ad6e6
2020-09-15 10:39:40 -07:00
Muthu Arivoli 9c364da9b9 Fix doc builds for bool kwargs (#44686)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/43669

The bool will still link to https://docs.python.org/3/library/functions.html#bool.
Tested using bmm:
![image](https://user-images.githubusercontent.com/16063114/93156438-2ad11080-f6d6-11ea-9b81-96e02ee68d90.png)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44686

Reviewed By: ngimel

Differential Revision: D23703823

Pulled By: mruberry

fbshipit-source-id: 7286afad084f5ab24a1254ad84e5d01907781c85
2020-09-15 10:34:58 -07:00
Yanli Zhao f5d231d593 move rebuild buckets from end of first iteration to beginning of second iteration (#44326)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44326

Part of relanding PR #41954, this refactoring is to move rebuild_buckets call from end of first iteration to beginning of second iteration
ghstack-source-id: 112011490

Test Plan: unit tests

Reviewed By: mrshenli

Differential Revision: D23583017

fbshipit-source-id: ef67f79437a820d9b5699b651803622418499a83
2020-09-15 09:51:33 -07:00
Vasiliy Kuznetsov 5f692a67db qat conv_fused.py: one more patch for forward compatibility (#44671)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44671

See comments inline - the FC between
https://github.com/pytorch/pytorch/pull/38478 and
https://github.com/pytorch/pytorch/pull/38820 was broken,
patching it.

Test Plan: Verified with customer hitting the issue that this fixes their issue.

Reviewed By: jerryzh168

Differential Revision: D23694029

fbshipit-source-id: a5e1733334e22305a111df750b190776889705d0
2020-09-15 09:43:29 -07:00
pinzhenx 72b5665c4f Upgrade oneDNN (mkl-dnn) to v1.6 (#44706)
Summary:
- Bump oneDNN (mkl-dnn) to 1.6 for bug fixes
    - Fixes https://github.com/pytorch/pytorch/issues/42446. RuntimeError: label is redefined for convolutions with large filter size on Intel AVX512
    - Implemented workaround for internal compiler error when building oneDNN with Microsoft Visual Studio 2019 (https://github.com/pytorch/pytorch/pull/43169)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44706

Reviewed By: ngimel

Differential Revision: D23705967

Pulled By: albanD

fbshipit-source-id: 65e8fecc52a76c9f3324403a8b60ffa8a8948bc6
2020-09-15 09:30:01 -07:00
Mike Ruberry 7036e91abd Revert D23323486: DPP Async Tracing
Test Plan: revert-hammer

Differential Revision:
D23323486 (https://github.com/pytorch/pytorch/commit/71673b31f932b671ace0b4d44a559891f770396e)

Original commit changeset: 4b6ca6c0e320

fbshipit-source-id: c6bd6d277aca070bef2de3522c2a60e23b4395ad
2020-09-15 01:19:23 -07:00
Michael Carilli 2435d941b1 Fix FP16 fastAtomicAdd for one case where tensor start address is not 32 bit aligned (#44642)
Summary:
For https://github.com/pytorch/pytorch/issues/44206 and https://github.com/pytorch/pytorch/issues/42218, I'd like to update trilinear interpolate backward and grid_sample backward to use `fastAtomicAdd`.

As a prelude, I spotted a UB risk in `fastAtomicAdd`.  I think existing code incurs a misaligned `__half2` atomicAdd when `index` is odd and `tensor` is not 32-bit aligned (`index % 2 == 1` and `(reinterpret_cast<std::uintptr_t>(tensor) % sizeof(__half2) == 1`). In this case we think we're `!low_bit` and go down the `!low_bit` code path, but in fact we are `low_bit`.  It appears the original [fastAtomicAdd PR](https://github.com/pytorch/pytorch/pull/21879#discussion_r295040377)'s discussion did not consider that case explicitly.

I wanted to push my tentative fix for discussion ASAP. jjsjann123 and mkolod as original authors of `fastAtomicAdd`. (I'm also curious why we need to `reinterpret_cast<std::uintptr_t>(tensor...` for the address modding, but that's minor.)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44642

Reviewed By: mruberry

Differential Revision: D23699820

Pulled By: ngimel

fbshipit-source-id: 0db57150715ebb45e6a1fb36897e46f00d61defd
2020-09-14 22:07:29 -07:00
Michael Carilli 2fd142a2ef Small clarification to amp gradient penalty example (#44667)
Summary:
requested by https://discuss.pytorch.org/t/what-is-the-correct-way-of-computing-a-grad-penalty-using-amp/95827/3

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44667

Reviewed By: mruberry

Differential Revision: D23692768

Pulled By: ngimel

fbshipit-source-id: 83c61b94e79ef9f86abed2cc066f188dce0c8456
2020-09-14 21:56:09 -07:00
Kyle Chen aedce773ed Deleted docker images for rocm 3.3 and rocm 3.5 (#44672)
Summary:
jeffdaily

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44672

Reviewed By: malfet

Differential Revision: D23694924

Pulled By: xw285cornell

fbshipit-source-id: 0066dc4b36c366588e1f309c82e7e1dc2ce8eec1
2020-09-14 21:50:41 -07:00
Vitaliy Chiley c71ce10cfc add dilation to transposeconv's _output_padding method (#43793)
Summary:
This PR adds dilation to _ConvTransposeNd._output_padding method and tests using a bunch of different sized inputs.

Fixes https://github.com/pytorch/pytorch/issues/14272

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43793

Reviewed By: zou3519

Differential Revision: D23493313

Pulled By: ezyang

fbshipit-source-id: bca605c428cbf3a97d3d24316d8d7fde4bddb307
2020-09-14 21:28:27 -07:00
peter ed862d3682 Split CUDA_NVCC_FLAGS by space (#44603)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/44599

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44603

Reviewed By: albanD

Differential Revision: D23692320

Pulled By: ezyang

fbshipit-source-id: 6a63d94ab8b88e7a82f9d65f03523d6ef639c754
2020-09-14 20:25:37 -07:00
Mike Ruberry 2c4b4aa81b Revert D23494065: Refactor CallbackManager as a nested class of RecordFunction.
Test Plan: revert-hammer

Differential Revision:
D23494065 (https://github.com/pytorch/pytorch/commit/63105fd5b192e2fe9310dd0445d73af79e3880b3)

Original commit changeset: 416d5bf6c942

fbshipit-source-id: 3b1ec928e3db0cc203bb63ec4db3da1584b9b884
2020-09-14 19:43:50 -07:00
Meghan Lele e7d782e724 [JIT] Add property support for ScriptModules (#42390)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/42390

**Summary**
This commit extends support for properties to include
ScriptModules.

**Test Plan**
This commit adds a unit test that has a ScriptModule with
a user-defined property.

`python test/test_jit_py3.py TestScriptPy3.test_module_properties`

Test Plan: Imported from OSS

Reviewed By: eellison, mannatsingh

Differential Revision: D22880298

Pulled By: SplitInfinity

fbshipit-source-id: 74f6cb80f716084339e2151ca25092b6341a1560
2020-09-14 18:49:21 -07:00
Louis Feng 63105fd5b1 Refactor CallbackManager as a nested class of RecordFunction. (#44645)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44645

Moved CallbackManager as a nested class of RecordFunction to allow private access to the call handles and context without exposing them publicly. It still hides the singleton instance of the CallbackManager inside record_function.cpp.

Test Plan: Unit tests.

Reviewed By: ilia-cher

Differential Revision: D23494065

fbshipit-source-id: 416d5bf6c9426e112877fbd233a6f4dff7bef455
2020-09-14 18:44:40 -07:00
Louis Feng 71673b31f9 DPP Async Tracing (#44252)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44252

Add tracing to DPP client. Because DPP requests are async, we need to be able to start a trace event in one thread and potentially end in a different thread. RecordFunction and LibgpumonObserver previously assume each trace event starts and finishes in the same thread. So they use a thread local context to track enter and exit call backs. Async events breaks this assumption. This change attaches the event context to the RecordFunction object so we do not need to use thread local context.

Test Plan:
Tested with dpp perf test and able to collect trace.

{F307824044}

Reviewed By: ilia-cher

Differential Revision: D23323486

fbshipit-source-id: 4b6ca6c0e32028fb38a476cd1f44c17a001fc03b
2020-09-14 18:43:14 -07:00
Guilherme Leobas e107ef5ca2 Add type annotations for torch.nn.utils.* (#43080)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/43013

Redo of gh-42954

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43080

Reviewed By: albanD

Differential Revision: D23681334

Pulled By: malfet

fbshipit-source-id: 20ec78aa3bfecb7acffc12eb89d3ad833024394c
2020-09-14 17:52:37 -07:00
Elias Ellison 551494b01d [JIT] Fix torch.tensor for empty multidimensional-typed lists (#44652)
Summary:
We were hitting an assert error when you passed in an empty `List[List[int]]` - this fixes that error by not recursing into 0-element tensors.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44652

Reviewed By: ZolotukhinM

Differential Revision: D23688247

Pulled By: eellison

fbshipit-source-id: d48ea24893044fae96bc39f76c0f1f9726eaf4c7
2020-09-14 17:28:23 -07:00
Kurt Mohler 2254e5d976 Add note comments to enforce nondeterministic alert documentation (#44140)
Summary:
This PR fulfills Ed's request (https://github.com/pytorch/pytorch/pull/41692#discussion_r473122076) for a strategy to keep the functions that have nondeterministic alerts fully documented.

Part of https://github.com/pytorch/pytorch/issues/15359

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44140

Reviewed By: colesbury

Differential Revision: D23644469

Pulled By: ezyang

fbshipit-source-id: 60936ccced13f071c620f7d25ef6dcbca338de7f
2020-09-14 16:48:22 -07:00
Facebook Community Bot a91c2be2a9 Automated submodule update: FBGEMM (#44647)
Summary:
This is an automated pull request to update the first-party submodule for [pytorch/FBGEMM](https://github.com/pytorch/FBGEMM).

New submodule commit: https://github.com/pytorch/FBGEMM/commit/1d710393d5b7588f5de3b83f51c22bbddf095229

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44647

Test Plan: Ensure that CI jobs succeed on GitHub before landing.

Reviewed By: dskhudia

Differential Revision: D23684528

fbshipit-source-id: 316ff2e448707a6e5a83248c9b22e58118bc8741
2020-09-14 16:43:59 -07:00
Mike Ruberry 686e281bcf Updates div to perform true division (#42907)
Summary:
This PR:

- updates div to perform true division
- makes torch.true_divide an alias of torch.div

This follows on work in previous PyTorch releases that first deprecated div performing "integer" or "floor" division, then prevented it by throwing a runtime error.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/42907

Reviewed By: ngimel

Differential Revision: D23622114

Pulled By: mruberry

fbshipit-source-id: 414c7e3c1a662a6c3c731ad99cc942507d843927
2020-09-14 15:50:38 -07:00
Jerry Zhang e594c30bc2 [quant][graphmode][fx] Support fp16 dynamic quantization for linear (#44582)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/44582

Test Plan:
test_quantize_fx.py

Imported from OSS

Reviewed By: vkuzo

Differential Revision: D23665974

fbshipit-source-id: 19ba6c61a9c77ef570b00614016506e9a2729f7c
2020-09-14 15:43:08 -07:00
BowenBao 43406e218a [ONNX] Update ONNX shape inference (#43929)
Summary:
* Support sequence type (de)serialization, enables onnx shape inference on sequence nodes.
* Fix shape inference with block input/output: e.g. Loop and If nodes.
* Fix bugs in symbolic discovered by coverage of onnx shape inference.
* Improve debuggability: added more jit logs. For simplicity, the default log level, when jit log is enabled, will not dump ir graphs.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43929

Reviewed By: albanD

Differential Revision: D23674604

Pulled By: bzinodev

fbshipit-source-id: ab6aacb16d0e3b9a4708845bce27c6d65e567ba7
2020-09-14 15:36:19 -07:00
Ivan Kobzarev 89aed1a933 [vulkan][op] avg_pool2d (#42675)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/42675

Test Plan: Imported from OSS

Reviewed By: AshkanAliabadi

Differential Revision: D22978765

Pulled By: IvanKobzarev

fbshipit-source-id: 64938d8965aeeb408dd5c40d688eca13fb7ebb8a
2020-09-14 15:07:34 -07:00
Ivan Kobzarev 8f327cd6c5 [vulkan][op] add.Scalar, mul.Scalar (#42674)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/42674

Test Plan: Imported from OSS

Reviewed By: AshkanAliabadi

Differential Revision: D22978763

Pulled By: IvanKobzarev

fbshipit-source-id: 9fd97d394205e3fa51992ee99d5bfafc33f75efa
2020-09-14 15:03:22 -07:00
Ksenija Stanojevic f7cfbac89b [ONNX] Update len symbolic (#43824)
Summary:
Update len symbolic.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43824

Reviewed By: izdeby

Differential Revision: D23575765

Pulled By: bzinodev

fbshipit-source-id: 0e5c8c8d4a5297f65e2dc43168993350f784c776
2020-09-14 15:00:44 -07:00
shubhambhokare1 da11d932bc [ONNX] Update arange op to support out argument (#43777)
Summary:
Update arange op to support out argument

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43777

Reviewed By: albanD

Differential Revision: D23674583

Pulled By: bzinodev

fbshipit-source-id: 6fb65e048c6b1a551569d4d2a33223522d2a960c
2020-09-14 14:56:17 -07:00
neginraoof 62ebad4ff9 [ONNX] Export new_empty and new_zeros (#43506)
Summary:
Adding symbolic to export new_empty and new_zeros

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43506

Reviewed By: houseroad

Differential Revision: D23674574

Pulled By: bzinodev

fbshipit-source-id: ecfcdbd4845fd3a3c6618a060129fbeee4df5dd7
2020-09-14 14:48:34 -07:00
Zafar d0a56cab07 [quant] Fixing the output shape for the linear (#44513)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/44513

Test Plan: Imported from OSS

Reviewed By: jerryzh168

Differential Revision: D23637508

Pulled By: z-a-f

fbshipit-source-id: d19d4c1b234b05e8d9813e864863d937b6c35bf5
2020-09-14 14:31:00 -07:00
Zafar 742654d1b6 [quant] ConvTranspose1d / ConvTranspose2d (#40371)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/40371

Test Plan: Imported from OSS

Reviewed By: vkuzo

Differential Revision: D22158981

Pulled By: z-a-f

fbshipit-source-id: defbf6fbe730a58d5b155dcb2460dd969797215c
2020-09-14 14:25:06 -07:00
Akihiro Nitta 84949672bf Fix exception chaining in test/ (#44193)
Summary:
## Motivation
This PR fixes https://github.com/pytorch/pytorch/issues/43770 and is the continuation of https://github.com/pytorch/pytorch/issues/43836.

## Description of the change
This PR fixes exception chaining only in files under `test/` where appropriate.
To fix exception chaining, I used either:
1. `raise new_exception from old_exception` where `new_exception` itself seems not descriptive enough to debug or `old_exception` delivers valuable information.
2. `raise new_exception from None` where raising both of `new_exception` and `old_exception` seems a bit noisy and redundant.

## List of lines containing `raise` in `except` clause:
I wrote [this simple script](https://gist.github.com/akihironitta/4223c1b32404b36c1b349d70c4c93b4d) using [ast](https://docs.python.org/3.8/library/ast.html#module-ast) to list lines where `raise`ing in `except` clause.

- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_cpp_extensions_aot.py#L16
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_jit.py#L2503
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/onnx/model_defs/word_language_model.py#L22
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/onnx/verify.py#L73
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/onnx/verify.py#L110
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/onnx/test_verify.py#L31
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_c10d.py#L255
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_c10d.py#L2992
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_c10d.py#L3025
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_c10d.py#L3712
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_distributed.py#L3180
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_distributed.py#L3198
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_data_parallel.py#L752
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/distributed/test_data_parallel.py#L776
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_type_hints.py#L151
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_jit_fuser.py#L771
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_jit_fuser.py#L773
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_dispatch.py#L105
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_distributions.py#L4738
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_nn.py#L9824
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_namedtensor.py#L843
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_jit_fuser_te.py#L875
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_jit_fuser_te.py#L877
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_dataloader.py#L31
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_dataloader.py#L43
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_dataloader.py#L365
- [x] https://github.com/pytorch/pytorch/blob/f8f35fddd4b2bd788953d6d6ccfadf690b73a3e8/test/test_dataloader.py#L391

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44193

Reviewed By: albanD

Differential Revision: D23681529

Pulled By: malfet

fbshipit-source-id: 7c2256ff17334625081137b35baeb816c1e53e0b
2020-09-14 14:20:16 -07:00
Alex Suhan a188dbdf3f Check for index-rank consistency in FunctionInliner (#44561)
Summary:
When caller / callee pairs are	inserted into the mapping, verify that
the arity of the buffer access is consistent with its declared rank.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44561

Test Plan: CI, test_tensorexpr --gtest_filter=TensorExprTest.DetectInlineRankMismatch

Reviewed By: albanD

Differential Revision: D23684342

Pulled By: asuhan

fbshipit-source-id: dd3a0cdd4c2492853fa68381468e0ec037136cab
2020-09-14 14:07:22 -07:00
Rong Rong b5dd6e3e61 split torch.testing._internal.* and add type checking for torch.testing._internal.common_cuda (#44575)
Summary:
First step to fix https://github.com/pytorch/pytorch/issues/42969.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44575

Reviewed By: malfet

Differential Revision: D23668740

Pulled By: walterddr

fbshipit-source-id: eeb3650b1780aaa5727b525b4e6182e1bc47a83f
2020-09-14 14:04:02 -07:00
mariosasko cfba33bde3 Fix the ELU formula in the docs (#43764)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/43389.

This PR replaces the old ELU formula from the docs that yields wrong results for negative alphas with the new one that fixes the issue and relies on the cases notation which makes the formula more straightforward.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/43764

Reviewed By: ailzhang

Differential Revision: D23425532

Pulled By: albanD

fbshipit-source-id: d0931996e5667897d926ba4fc7a8cc66e8a66837
2020-09-14 14:01:56 -07:00
Zafar 9d4943daaf [quant] conv_transpose1d / conv_transpose2d (#40370)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/40370

Test Plan: Imported from OSS

Reviewed By: vkuzo

Differential Revision: D22158979

Pulled By: z-a-f

fbshipit-source-id: f5cb812c9953efa7608f06cf0188de447f73f358
2020-09-14 13:45:28 -07:00
Rong Rong ecac8294a6 enable type checking for torch._classes (#44576)
Summary:
Fix https://github.com/pytorch/pytorch/issues/42980

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44576

Reviewed By: malfet

Differential Revision: D23668741

Pulled By: walterddr

fbshipit-source-id: 4201ea3187a40051ebff53d28c8e571ea1a61126
2020-09-14 13:26:46 -07:00
Raghavan Raman ad7a2eb1c9 Simplify nested Min and Max patterns. (#44142)
Summary:
Improve simplification of nested Min and Max patterns.

Specifically, handles the following pattern simplications:
  * `Max(A, Max(A, Const)) => Max(A, Const)`
  * `Max(Min(A, B), Min(A, C)) => Min(A, Max(B, C))`
  * `Max(Const, Max(A, OtherConst) => Max(A, Max(Const, OtherConst))`
     - This case can have an arbitrarily long chain of Max ops. For example: `Max(5, Max(x, Max(y, Max(z, 8)))) => Max(Max(Max(x, 8), y), z)`

Similarly, for the case of Min as well.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44142

Reviewed By: albanD

Differential Revision: D23644486

Pulled By: navahgar

fbshipit-source-id: 42bd241e6c2af820566744c8494e5dee172107f4
2020-09-14 13:24:46 -07:00
Heitor Schueroff de Souza 199435af90 Update median doc to note return value of even-sized input (#44562)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44562

Add a note that torch.median returns the smaller of the two middle elements for even-sized input and refer user to torch.quantile for the mean of the middle values.

fixes https://github.com/pytorch/pytorch/issues/39520

Test Plan: Imported from OSS

Reviewed By: albanD

Differential Revision: D23657208

Pulled By: heitorschueroff

fbshipit-source-id: 2747aa652d1e7f10229d9299b089295aeae092c2
2020-09-14 13:18:33 -07:00
Bram Wasti a475613d1d [static runtime] Swap to out-variant compatible nodes (#44127)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/44127

Test Plan: Imported from OSS

Reviewed By: hlu1

Differential Revision: D23604306

Pulled By: bwasti

fbshipit-source-id: 18ccfb9b466b822e28130be3d5c4fae36c76820b
2020-09-14 12:38:25 -07:00
Elias Ellison 856510c96d [JIT] Dont optimize shape info in batch_mm (#44565)
Summary:
We run remove profile nodes and specialize types before batch_mm, so we cannot run peepholes on the type information of tensors since these properties have not been guarded to be guaranteed to be correct.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44565

Reviewed By: albanD

Differential Revision: D23661538

Pulled By: eellison

fbshipit-source-id: 0dd23a65714f047f49b4db4ec582b21870925fe1
2020-09-14 12:34:20 -07:00
Marcin Juszkiewicz e261e0953e Fix centos8 gcc (#44644)
Summary:
Fixes https://github.com/pytorch/pytorch/issues/44198 properly this time

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44644

Reviewed By: albanD

Differential Revision: D23684909

Pulled By: malfet

fbshipit-source-id: cea6f6e2ae28138f6b93a6513d1abd36d14ae573
2020-09-14 12:28:09 -07:00
Yi Wang ace81b6794 Remove an extra empty line in the warning comments. (#44622)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44622

Remove an extra empty line in the warning comments.Remove an extra empty line.

Test Plan: N/A

Reviewed By: rohan-varma

Differential Revision: D23674070

fbshipit-source-id: 4ee570590c66a72fb808e9ee034fb773b833efcd
2020-09-14 11:15:35 -07:00
Heitor Schueroff de Souza 21a09ba94d Fix lerp.cu bug when given discontiguous out tensor (#44559)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44559

Please refer to the discussion at the bottom of https://github.com/pytorch/pytorch/pull/43541 about the bug.

Test Plan: Imported from OSS

Reviewed By: albanD

Differential Revision: D23655403

Pulled By: heitorschueroff

fbshipit-source-id: 10e4ce5c2fe7bf6e95bcfac4033202430292b03f
2020-09-14 11:03:02 -07:00
Natalia Gimelshein 95a69a7d09 adds list_gpu_processes function (#44616)
Summary:
per title, to make it easier to track the creation of stray contexts:
```
python -c "import torch; a=torch.randn(1, device='cuda'); print(torch.cuda.memory.list_gpu_processes(0)); print(torch.cuda.memory.list_gpu_processes(1))"
GPU:0
process      79749 uses      601.000 MB GPU memory
GPU:1
no processes are running
```

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44616

Reviewed By: mruberry

Differential Revision: D23675739

Pulled By: ngimel

fbshipit-source-id: ffa14cad9d7144e883de13b1c2c6817bd432f53a
2020-09-14 09:54:32 -07:00
Rong Rong 105132b891 Move ONNX circle ci build to torch and remove all caffe2 CI job/workflows (#44595)
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/44595

Reviewed By: seemethere

Differential Revision: D23670280

Pulled By: walterddr

fbshipit-source-id: b32633912f6c8b4606be36b90f901e636567b355
2020-09-14 09:50:13 -07:00
Thomas Viehmann bd257a17a1 Add HIP/ROCm version to collect_env.py (#44106)
Summary:
This adds HIP version info to the `collect_env.py` output.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44106

Reviewed By: VitalyFedyunin

Differential Revision: D23652341

Pulled By: zou3519

fbshipit-source-id: a1f5bce8da7ad27a1277a95885934293d0fd43c5
2020-09-14 09:19:18 -07:00
Jeremy Lilley 7040a070e3 [torch] Minor: Avoid ostreamstring in Operator's canonicalSchemaString() (#44442)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/44442

I noticed lock contention on startup as lookupByLiteral() was
calling registerPendingOperators() - some calls were holding the
lock for 10+ ms, as operators were being registered.

canonicalSchemaString() was using ostreamstring, which isn't typically
particularly fast (partly because of c++ spec locale requirements).
If we repalce with regular c++ string appends, it's somewhat faster
(which isn't hard when comparing with stringstream; albeit a bit
more codegen)

Over the first minute or so, this cuts out 1.4 seconds under the
OperatorRegistry lock (as part of registerPendingOperators) in the
first couple minutes of run time (mostly front-loaded) when running
sync sgd.

As an example, before:
   registerPendingOperators 12688 usec for 2449 operators
After:
   registerPendingOperators 6853 usec for 2449 operators
ghstack-source-id: 111862971

Test Plan: buck test mode/dev-nosan caffe2/test/cpp/...

Reviewed By: ailzhang

Differential Revision: D23614515

fbshipit-source-id: e712f9dac5bca0b1876e11fb8f0850402f03873a
2020-09-14 08:24:16 -07:00
kshitij12345 c68a99bd61 [numpy] Add torch.exp2 (#44184)
Summary:
Reference https://github.com/pytorch/pytorch/issues/42515

TODO
* [x] Add tests
* [x] Add docs

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44184

Reviewed By: ngimel

Differential Revision: D23674237

Pulled By: mruberry

fbshipit-source-id: 7f4fb1900fad3051cd7fc9d3d7f6d985c5fb093c
2020-09-14 04:05:37 -07:00
Facebook Community Bot 870f647040 Automated submodule update: FBGEMM (#44581)
Summary:
This is an automated pull request to update the first-party submodule for [pytorch/FBGEMM](https://github.com/pytorch/FBGEMM).

New submodule commit: https://github.com/pytorch/FBGEMM/commit/0725301da504c6ddf1633720ca08d17a8e03914a

Pull Request resolved: https://github.com/pytorch/pytorch/pull/44581

Test Plan: Ensure that CI jobs succeed on GitHub before landing.

Reviewed By: dskhudia, VitalyFedyunin

Differential Revision: D23665173

fbshipit-source-id: 03cee22335eef0517e561827795bbe2036942ea0
2020-09-13 21:26:56 -07:00