Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20924

I found a python3 bug for deserializing caffe2 code. The exception thrown is Unicode related error instead of just decode error, and we need to catch that as well

Reviewed By: ipiszy

Differential Revision: D15293221

fbshipit-source-id: 29820800d1b4cbe5bf3f5a189fe2023e655d0508

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21763

Custom __getattr__ functions can only raise AttributeError. This code throwed NotImplementedError which caused upstream troubles when hasattr() was called.

Differential Revision: D15815176

fbshipit-source-id: 0982e2382de4578d3fc05c5d2a63f624d6b4765e

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21652

Diff fixes issue of empty ROIs for convTranspose

Issue StackTrace: P65374505

Reviewed By: jerryzh168

Differential Revision: D15766739

fbshipit-source-id: 39cf8feca66b6aae22ff4ec5c1b6a4e3f20f378d

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21647
ghimport-source-id: 1db4ec31f047f7854a39c28e2b38918dc6b44f42

Differential Revision: D15804425

Pulled By: zou3519

fbshipit-source-id: 575cc3de09287efe75e7052df129626748208d0d

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21750
ghimport-source-id: 4792aa5ccab7e4b54c21f23d0b78802f85bbeb8d

Differential Revision: D15819367

Pulled By: ezyang

fbshipit-source-id: db91ee727c66469ac78e59b3662f29db53a916bc

Summary:
fixes https://github.com/pytorch/hub/issues/29
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21685

Differential Revision: D15817774

Pulled By: ailzhang

fbshipit-source-id: d2f615e5d431186d45a21d8300fb9ba3c37b246c

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21446

this is used for easier tracing of iter id when looking at trace diagram

Reviewed By: ilia-cher

Differential Revision: D15628950

fbshipit-source-id: ee75b3bdb14a36abc18c7bddc49d8ec9789b724d

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21141

Differential Revision: D15808354

Pulled By: ZolotukhinM

fbshipit-source-id: 16d938fd5acafb445a0c433cabc9a55cab563165

Summary:
```
The stride calculation using OffsetCalculator performs poorly with
MAX_DIMS=25. This reduces MAX_DIMS (after coalescing) to 16 on ROCm.
I think it's unlikely that anyone will exceed this limit. If they do,
we can add additional specializations for ROCm with more dimensions.
```

I'm not sure about the underlying cause. With MAX_DIM=25, the add kernel's params
is ~648 bytes vs. ~424 bytes with MAX_DIM=16. The kernel instruction footprint is
bigger too, but most of these instructions are never executed and most kernel parameters
are never loaded because the typical dimensionality is much smaller.

Mini benchmark here:
https://gist.github.com/colesbury/1e917ae6a0ca9d24712121b92fed4c8f

(broadcasting operations are much faster)

cc iotamudelta
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21754

Reviewed By: bddppq

Differential Revision: D15811906

Pulled By: colesbury

fbshipit-source-id: 063f92c083d26e2ef2edc98df7ff0400f9432b9d

Summary:
alloctor -> allocator
excutable -> executable
excution -> execution
foward -> forward
initiaize -> initialize
paralell -> parallel
preprocesor -> preprocessor
tranpose -> transpose
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21665

Differential Revision: D15806155

Pulled By: soumith

fbshipit-source-id: d92b21ec8650a2b32f05faf9af0b7d2b073e992c

Summary:
Currently multihead attention for half type is broken
```
  File "/home/ngimel/pytorch/torch/nn/functional.py", line 3279, in multi_head_attention_forward
    attn_output = torch.bmm(attn_output_weights, v)
RuntimeError: Expected object of scalar type Float but got scalar type Half for argument https://github.com/pytorch/pytorch/issues/2 'mat2'
```
because softmax converts half inputs into fp32 inputs. This is unnecessary - all the computations in softmax will be done in fp32 anyway, and the results need to be converted into fp16 for the subsequent batch matrix multiply, so nothing is gained by writing them out in fp32. This PR gets rid of type casting in softmax, so that half works.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21658

Differential Revision: D15807487

Pulled By: zhangguanheng66

fbshipit-source-id: 4709ec71a36383d0d35a8f01021e12e22b94992d

Summary:
In this PR, we use `expect` to fill in the token for pytorchbot when doing `git push`, so that we don't need to save the token in the git remote URL.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20459

Differential Revision: D15811676

Pulled By: yf225

fbshipit-source-id: cd3b780da05d202305f76878e55c3435590f15a8

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21755
ghimport-source-id: dfb53759024d9ba9d104fdb2a8151ab996e55234

Differential Revision: D15811172

Pulled By: ezyang

fbshipit-source-id: c8c7c1c15277d8fe8cc513e20af449257d7ff15c

Summary:
Fix an obvious bug.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21669

Reviewed By: zrphercule

Differential Revision: D15806614

Pulled By: houseroad

fbshipit-source-id: d0f6e934252e0057f3dbcc7f160236ee6f4497ac

Summary:
Fixing reported [bug](https://github.com/pytorch/pytorch/issues/20322)
The issue was related to not checking the dimensions of source vs destination tensors.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21617

Differential Revision: D15749963

Pulled By: izdeby

fbshipit-source-id: acff114c729fd9c0a9a51325e0ebd8b42e1f2fc1

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21742

Add error message to NotImplementedError so we know which function it is about.

Reviewed By: bddppq

Differential Revision: D15806379

fbshipit-source-id: 14eab9d03aa5b44ab95c5caeadc0e01d51f22188

Summary:
Add docs for TransformerEncoder and TransformerDecoder, plus minor edits.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21746

Differential Revision: D15807498

Pulled By: zhangguanheng66

fbshipit-source-id: 388efb5821c4c3d25865cecea70902e9b2bf5d15

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21751

Pulled By: driazati

Differential Revision: D15809091

fbshipit-source-id: 3cc96e632a7b89b4d86d68d2a76021d971447e12

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21747

Differential Revision: D15808189

Pulled By: ezyang

fbshipit-source-id: 5413eaaa901be098c6bad135f702ba103bc79d6c

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21739

Added event and event_counter columns for PyTorch/Caffe2 API usage metrics

Reviewed By: dzhulgakov

Differential Revision: D15119119

fbshipit-source-id: a71010bd659109a8e4f3a8bad84b22c1d15dc528

Summary:
When converting pixel_shuffle to reshape + transpose + reshape, the first reshape should
be:
[N, C * r^2, H, W] => [N, C, r, r, H, W]
in order to match pytorch's implementation (see ATen PixelShuffle.cpp).

This previously wasn't caught by the test case, since it uses C = r = 4. Updated test case to
have C = 2, r = 4.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21486

Reviewed By: houseroad

Differential Revision: D15700945

Pulled By: houseroad

fbshipit-source-id: 47019691fdc20e152e867c7f6fd57da104a12948

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21581
ghimport-source-id: 6a65d73694b17611d6ad45db0b39b86c318a68c7

Differential Revision: D15736495

Pulled By: ilia-cher

fbshipit-source-id: 6b9109ad3611ff3c8b1a37796e9149bef0c2ad36

Summary:
](https://our.intern.facebook.com/intern/diff/15800009/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21725

Pulled By: driazati

Differential Revision: D15800009

fbshipit-source-id: 5409c213161e3f2031710933897b85872aad2a83

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20480
ghimport-source-id: c710f897c4c9b9616fc3dd76d80b4845aea43a1f

Differential Revision: D15333692

Pulled By: ilia-cher

fbshipit-source-id: 61e476dd5c737fe144e3aec000d8ebb11fbc0547

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21635

Differential Revision: D15768487

Pulled By: ezyang

fbshipit-source-id: 85e1d883aded0f4d3ac5100719df335f5a337fc5

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21718

adding a detection method on whether the package is built for AMD.

Reviewed By: bddppq

Differential Revision: D15795893

fbshipit-source-id: 91a21ee76b2273b1032507bdebe57e016717181d

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21724

Differential Revision: D15799149

Pulled By: bddppq

fbshipit-source-id: c72689e73470f2ca145556a2ac8cb34e36e341ef

Summary:
**Closes:** Confusing documentation with distributions.Categorical about logits https://github.com/pytorch/pytorch/issues/16291

**Solution**: Changes documentation on the Categorical distribution from `log probabilities` to `event log-odds`. This makes should reduce confusion as raised by this issue, and is consistent with other distributions such as `torch.Binomial`.

More than happy to make any other changes if they fit :).
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21707

Differential Revision: D15799181

Pulled By: soumith

fbshipit-source-id: f11acca7a5c130102a3ff6674640235ee5aa69bf

Summary:
- [x] Add tests after https://github.com/pytorch/pytorch/pull/20256 is merged

- Support exporting ScriptModule with inputs/outputs of arbitrarily constructed tuples.

- Moved the assigning of output shapes to after graph conversion to ONNX is completed. By then all tuples in the IR has already been lowered by the pass ```_jit_pass_lower_all_tuples```. If assigning output shapes is required to happen before that, we'll need to hand parse the tuple structures in the graph, and repeat the same logic in ```_jit_pass_lower_all_tuples```. Handling inputs is easier because all tuple information is encoded within the input tensor type.

- Swap the order of ```_jit_pass_lower_all_tuples``` and ```_jit_pass_erase_number_types```. Ops like ```prim::TupleIndex``` relies on index being a scalar. ```_jit_pass_erase_number_types``` will convert these kind of scalars to tensors.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20784

Reviewed By: zrphercule

Differential Revision: D15484171

Pulled By: houseroad

fbshipit-source-id: 4767a84038244c929f5662758047af6cb92228d3

Summary:
…ngular_solve

Changelog:
- Iterate over mini batches of 65535 matrices (maximum)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21689

Differential Revision: D15800254

Pulled By: soumith

fbshipit-source-id: c743ff13f1ba25d26874429d44e41a3c0ed21d6a

Summary:
### mkldnn backward ops list:
 - [ ] \(https://github.com/pytorch/pytorch/pull/20567) Add aten mkldnn conv2d backward operator 💛
 - [ ] \(https://github.com/pytorch/pytorch/pull/20570) Add aten mkldnn backward ops: relu, linear and reshape 💛
 - [ ] \(https://github.com/pytorch/pytorch/pull/20571) Add aten mkldnn backward ops: max_pool2d, avg_pool2d and adaptive_avg_poo2d 💛
 - [ ] \(https://github.com/pytorch/pytorch/pull/20572) Add aten mkldnn batchnorm backward operator 💛
 - [ ] \(https://github.com/pytorch/pytorch/pull/20573) Add aten mkldnn zero_ operator:yellow_heart:
 - [ ] \(https://github.com/pytorch/pytorch/pull/20575) Add mkldnn mul operator 💛
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20575

Differential Revision: D15799529

Pulled By: bddppq

fbshipit-source-id: 4887d8ef1a0e316ad9db199b657d9481fc13e486

Differential Revision:
D15769066

Original commit changeset: 5853e0360581

fbshipit-source-id: ac6fa8429136abf4c7835919009f936eea11ea7b

Summary:
This renames the CMake `caffe2` target to `torch`, as well as renaming `caffe2_gpu` to `torch_gpu` (and likewise for other gpu target variants).  Many intermediate variables that don't manifest as artifacts of the build remain for now with the "caffe2" name; a complete purge of `caffe2` from CMake variable names is beyond the scope of this PR.

The shell `libtorch` library that had been introduced as a stopgap in https://github.com/pytorch/pytorch/issues/17783 is again flattened in this PR.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/20774

Differential Revision: D15769965

Pulled By: kostmo

fbshipit-source-id: b86e8c410099f90be0468e30176207d3ad40c821

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21719

Differential Revision: D15797628

Pulled By: Krovatkin

fbshipit-source-id: 87742bdde0b05aff4341ababb1f55c51991768ec

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21695

Differential Revision: D15795716

Pulled By: Krovatkin

fbshipit-source-id: e14a44210ea4312a247157a6681fce449e40f779

Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/21615

Differential Revision: D15793434

Pulled By: Krovatkin

fbshipit-source-id: d89f1bf61ea57a1e3b75f8e2b200c27beb8b46cf

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21565
ghimport-source-id: d1fe735fb7821eadc59116fb921d8fe39a49f818

Reviewed By: driazati

Differential Revision: D15729503

Pulled By: zdevito

fbshipit-source-id: fabb678f040d21fae7545e3b2be1d098e24c544e

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21660
ghimport-source-id: f9a11b2748f49042ee636755358d79c547aa249e

Reviewed By: suo

Differential Revision: D15770237

Pulled By: zdevito

fbshipit-source-id: 41fa8577028eef247bc545635cd93192a0b19db4

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21515
ghimport-source-id: 7898a68791db2b5050164ab01d6ca6991e05746d

Reviewed By: suo

Differential Revision: D15719981

Pulled By: zdevito

fbshipit-source-id: 42cf26cf6541bcdf95f1343da3b9228fe2c229da

Summary:
Class member annotations can be marked with `Final[T]` instead of adding them to `__constants__`. `Final` comes from the `typing_extensions` module (which will be used if it is present). If not, the polyfill from `_jit_internal` is exposed as `torch.jit.Final` for users that don't want to install `typing_extensions`.

This keeps around `__constants__` since a lot of code is still using it, but in documentation follow ups we should change the examples to all to use `Final`.

TODO: install typing_extensions on CI, move tests to a Python3 only file when #21489 lands
](https://our.intern.facebook.com/intern/diff/15746274/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21603

Pulled By: driazati

Differential Revision: D15746274

fbshipit-source-id: d2c9b5643b4abba069b130c26fd42714c906ffac