Sebastian Messmer [Fri, 22 Mar 2019 21:05:50 +0000 (14:05 -0700)]
Avoid refcount when looking up dispatch key
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18294
Reviewed By: ezyang
Differential Revision:
D14512979
fbshipit-source-id:
45e548974f06184c375c2bb8339e3049a4ebd880
Jiakai Liu [Fri, 22 Mar 2019 21:01:41 +0000 (14:01 -0700)]
Fix DCHECK to handle dangling else (#18295)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18295
Replace "if (false)" with "while (false)" which fixes potential dangling else issue as shown in added test case.
Reviewed By: ezyang
Differential Revision:
D14569608
fbshipit-source-id:
407052db9182ce27b7a59841e90fa50d3eca262e
Natalia Gimelshein [Fri, 22 Mar 2019 20:48:59 +0000 (13:48 -0700)]
Allow fusion of float function arguments (#18087)
Summary:
so that functions like `def fn(x, p:float)` can be fused. Fixes #9940 and #11186. Fuses only float (not integer) arguments to simplify assembling arguments for fusion launch.
CPU fusion is disabled in CI and this won't be tested, but I tested it locally.
cc t-vi, apaszke
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18087
Differential Revision:
D14581206
Pulled By: wanchaol
fbshipit-source-id:
ccb0cf79b1751706f9b2cdf1715115eae5a39fb6
Thomas Viehmann [Fri, 22 Mar 2019 20:31:37 +0000 (13:31 -0700)]
Fix error reporting in NVRTC use of the fuser (#18327)
Summary:
Two functions were not directed ad NVRTC.
It's a bit hard to test this, as the fuser usually produces correct code - unless I try to hack on it. :)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18327
Differential Revision:
D14579285
Pulled By: soumith
fbshipit-source-id:
1be7ba461cc473d514ba619507742a47d4d7c97e
Ailing Zhang [Fri, 22 Mar 2019 20:22:52 +0000 (13:22 -0700)]
Using sqrt for better precision in cosine_similarity (#18250)
Summary:
address comment in #18168 .
Testing in CI...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18250
Differential Revision:
D14568601
Pulled By: ailzhang
fbshipit-source-id:
39fbbdb08743b53fa665c7e88e4750cbe0976ec7
Jianyu Huang [Fri, 22 Mar 2019 19:28:04 +0000 (12:28 -0700)]
Fix alignment issues for Fake BFP16 fp32 -> bfp16 rounding routines (#18321)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18321
As title.
Reviewed By: jspark1105
Differential Revision:
D14575512
fbshipit-source-id:
0e33cdab54b1aef8b67f0b4c366692c5dbdf631d
Dmytro Dzhulgakov [Fri, 22 Mar 2019 19:10:19 +0000 (12:10 -0700)]
Untangle internal build python and cpp dependencies
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18326
Reviewed By: ezyang
Differential Revision:
D14576294
fbshipit-source-id:
186ce1e3d026d962b7386f861eddf093f583a878
Alexander Sidorov [Fri, 22 Mar 2019 18:49:04 +0000 (11:49 -0700)]
Caffe2: crash op (#18207)
Summary:
this is handy when testing various core dump related
things. If in the future we want to unit test our future gdb debugger
extensions, we can use this op to generate a core dump for us within a
unit test.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18207
Differential Revision:
D14482186
Pulled By: salexspb
fbshipit-source-id:
39a9fffbdd4bd083597f544d1c783a82cf023a89
Duc Ngo [Fri, 22 Mar 2019 18:14:40 +0000 (11:14 -0700)]
caffe2 - Util to cleanup external inputs and outputs from a NetDef (#18194)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18194
Add a util method to cleanup external inputs and outputs from a NetDef
The following conditions will be met after the modification
- No duplicate external inputs
- No duplicate external outputs
- Going through list of ops in order, all op inputs must be outputs
from other ops, or registered as external inputs.
- All external outputs must be outputs of some operators.
Reviewed By: ZolotukhinM
Differential Revision:
D14528589
fbshipit-source-id:
c8d82fda1946aa3696abcbec869a4a8bb22f09b6
Dmytro Dzhulgakov [Fri, 22 Mar 2019 18:11:16 +0000 (11:11 -0700)]
End to end hack to call server side Caffe2 ops (#18267)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18267
Motivation: we don't actually want to use it for real under any circumstances. This is an idea to unblock our internal progress and parallelize workstreams. We can easily define schemas for all ops in question and implement forwarding to C2 ops which is NOT going to be performant. Then several things can be happening in parallel:
* move code of ops outside of C2 ops that depend on protobuf into c10
* development of optimization/fusion passes
* building python-level wrappers with clean API
* improving perf
This demonstrates, Relu, quant, dequant. It seems to cover all use cases necessary (maybe except weights prepacking). Ideally I'd demonstrate Conv, but will get to it later in a separate PR (contributions welcomed)
Reviewed By: ezyang
Differential Revision:
D14531232
fbshipit-source-id:
4cd4a71ae0cb373c6c0e81f965c442b82a1b4069
Bilge Acun [Fri, 22 Mar 2019 16:51:27 +0000 (09:51 -0700)]
Optimize MomentumSGDUpdate maximum block size and make it templated
Summary: Removing the maximum number of blocks limit from the operator and making the nesterov parameter templated to remove branching.
Reviewed By: BIT-silence
Differential Revision:
D14567003
fbshipit-source-id:
394c2039ee214adc6ccd2e562e4e9563d307131f
Edward Yang [Fri, 22 Mar 2019 14:46:50 +0000 (07:46 -0700)]
Add test for #17271 (torch.exp incorrect for 2**31 size tensor) (#18292)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18292
ghimport-source-id:
a3e96584db0eef7b6202a1211808f9f6e59dd529
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18292 Add test for #17271 (torch.exp incorrect for 2**31 size tensor)**
* #18291 Correctly call superclass setUp in TestCase subclasses.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14567642
fbshipit-source-id:
c60ee7597a86f5d2c5c0b72cb106f17815950427
Edward Yang [Fri, 22 Mar 2019 14:43:40 +0000 (07:43 -0700)]
Correctly call superclass setUp in TestCase subclasses. (#18291)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18291
ghimport-source-id:
d6e95e899bd320407967df41435801e54864ba62
Stack from [ghstack](https://github.com/ezyang/ghstack):
* #18292 Add test for #17271 (torch.exp incorrect for 2**31 size tensor)
* **#18291 Correctly call superclass setUp in TestCase subclasses.**
This makes PYTORCH_TEST_SKIP_FAST work correctly for more
tests, reducing the wasted testing effort on our slow_test job.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14567643
fbshipit-source-id:
40cf1d6556e0dd0a0550ff3d9ffed8b6000f8191
Gerard Goossen [Fri, 22 Mar 2019 13:33:24 +0000 (06:33 -0700)]
Verify def before infer fensor (#18129)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18129
A lot of tensor interference function assume the operator passes the schema.
So call Verity to make sure this is actually the case.
Created diff before to add checking in Concat (https://github.com/pytorch/pytorch/pull/17110), but I encountered lot more places where this is assumed (for example ElementwiseOpShapeInference)
Reviewed By: mdschatz
Differential Revision:
D14503933
fbshipit-source-id:
cf0097b8c3e4beb1cded6b61e092a6adee4b8fcb
Jongsoo Park [Fri, 22 Mar 2019 07:49:11 +0000 (00:49 -0700)]
add more Python interface functions to make quantization simpler (#18246)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18246
Simplifies histogram collection and quantization process.
Histogram collection before this diff was something like this
```
from caffe2.quantization.server import dnnlowp_pybind11
...
dnnlowp_pybind11.ObserveHistogramOfOutput(hist_file)
for ...
workspace.RunNet(predict_net)
dnnlowp_pybind11.ClearNetObservers() # This is to trigger Stop function in the observer to dump out histogram file but this can have unintended consequence of also clearing all the other useful observers we attached
```
After this diff we can
```
workspace.CreateNet(predict_net) # Note we need to create net to have a net to attach observer
histogram_observer = dnnlowp_pybind11.AddHistogramObserver(predic_net, hist_file)
for ...
workspace.RunNet(predict_net)
predict_net.RemoveObserver(histogram_observer)
```
Choosing quantization parameters of weights before this diff was something like this
```
dnnlowp_pybind11.ObserveHistogramOfOutput(weight_hist_file)
workspace.RunNetOnce(init_net)
dnnlowp_pybind11.ClearNetObservers() # Has same issue as the histogram collection example above
dnnlowp_pybind11.RegisterQuantizationParamsWithHistogram(
weight_hist_file, is_weight=True, qparams_output_file_name=qparams_file
)
workspace.CreateNet(init_net, overwrite=True)
dnnlowp_pybind11.ClearNetObservers()
logger.info("Loading quantization params from {}".format(qparams_file))
blobs_to_qparams = {}
with open(qparams_file) as f:
lines = f.readlines()
for line in lines:
op_id, op_type, output_id, tensor_name, mini, maxi, scale, zero_point, precision = (
line.split()
)
op_id = int(op_id)
output_id = int(output_id)
op = net.Proto().op[op_id]
if op_type != op.type or op.output[output_id] != tensor_name:
print(
"Corrupt qparams file {} {} {} {} {}".format(
qparams_file, op_type, op.type, op.output[output_id], tensor_name
)
)
blobs_to_qparams[tensor_name] = QuantizationParam(float(scale), int(zero_point))
```
After this diff this can be simplified to
```
blobs_to_qparams = {}
for op in init_net.Proto().op:
for output in op.output:
scale, zero_point = dnnlowp_pybind11.ChooseQuantizationParams(output)
blobs_to_qparams[output] = QuantizationParam(scale, zero_point)
```
Reviewed By: dskhudia
Differential Revision:
D14544694
fbshipit-source-id:
4fd06cd63256201e2e9d15c39f503138d1be53c2
Weiyi Zheng [Fri, 22 Mar 2019 07:08:50 +0000 (00:08 -0700)]
add fbgemm fp16 (fbfcpacked) support, add global_init_net in predictor_export_meta (#18257)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18257
support adding op in global_init_net. because pred_init_net is per thread, and just doesn't cut it.
Reviewed By: jspark1105
Differential Revision:
D14552695
fbshipit-source-id:
53dd44c84ad019019ab9f35fc04d076b7f941ddc
Lu Fang [Fri, 22 Mar 2019 07:07:57 +0000 (00:07 -0700)]
update of fbcode/onnx to
c05f2ae412daf8fd64136ca354b97ccf73e0ea6c (#18285)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18285
Previous import was
96c58ceeacf0f2b73d752e413e4fd78787a12da3
Included changes:
- **[
c05f2ae4](https://github.com/onnx/onnx/commit/
c05f2ae4)**: update both core and ml docs (#1879) <Lu Fang>
- **[
f895279b](https://github.com/onnx/onnx/commit/
f895279b)**: fix the problems introduced in previous PRs in operator registration (#1878) <Lu Fang>
- **[
f6f80657](https://github.com/onnx/onnx/commit/
f6f80657)**: Skip the schema check on ops in non-standard domain (#1876) <Lu Fang>
- **[
8c8be722](https://github.com/onnx/onnx/commit/
8c8be722)**: Introduce Function Body Helper (#1868) <Sherlock>
- **[
b605eafb](https://github.com/onnx/onnx/commit/
b605eafb)**: Support down sampling for Upsample with scales < 1. (#1773) <Ke Zhang>
- **[
47f7aa71](https://github.com/onnx/onnx/commit/
47f7aa71)**: Remove scaledtanh (#1866) <Ashwini Khade>
- **[
4dfc56de](https://github.com/onnx/onnx/commit/
4dfc56de)**: Add Ceil support for Max and Average Pooling (#1860) <Lara Haidar>
- **[
552a8efc](https://github.com/onnx/onnx/commit/
552a8efc)**: Add testcase generator for functions (#1862) <Raymond Yang>
- **[
fdb978a5](https://github.com/onnx/onnx/commit/
fdb978a5)**: Promote Thresholded Relu Op (#1856) <Ashwini Khade>
- **[
ce332628](https://github.com/onnx/onnx/commit/
ce332628)**: Update Slice with dynamic input & optional input steps (#1836) <Bowen Bao>
- **[
3a9a8787](https://github.com/onnx/onnx/commit/
3a9a8787)**: Merge function into opschema (#1834) <Raymond Yang>
- **[
3dbf8fe9](https://github.com/onnx/onnx/commit/
3dbf8fe9)**: Handle string comparision represented as np.objects (#1851) <Dmitri Smirnov>
- **[
3b0d3bb2](https://github.com/onnx/onnx/commit/
3b0d3bb2)**: remove global variable in header file (#1850) <Lu Fang>
- **[
1cca8733](https://github.com/onnx/onnx/commit/
1cca8733)**: bump the version for drop out - fix the issue that the version was not bumped when changing its type constraint declaration. (#1848) <Ke Zhang>
- **[
1ec81bc6](https://github.com/onnx/onnx/commit/
1ec81bc6)**: Change TopK operator to allow dynamic 'k' (#1829) <Hariharan Seshadri>
- **[
a89a4a16](https://github.com/onnx/onnx/commit/
a89a4a16)**: Remove exp op: Affine, ImageScaler,ParametricSoftplus, Crop. (#1832) <Ke Zhang>
Reviewed By: yinghai
Differential Revision:
D14566202
fbshipit-source-id:
b1e5912ae6887e2865fc628363071e2b9938dfa4
David Riazati [Fri, 22 Mar 2019 03:15:38 +0000 (20:15 -0700)]
Cleanup TorchScript rst docs (#18234)
Summary:
* Adds more headers for easier scanning
* Adds some line breaks so things are displayed correctly
* Minor copy/spelling stuff
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18234
Reviewed By: ezyang
Differential Revision:
D14567737
Pulled By: driazati
fbshipit-source-id:
046d991f7aab8e00e9887edb745968cb79a29441
Junjie Bai [Thu, 21 Mar 2019 23:24:45 +0000 (16:24 -0700)]
Replace the remaining usages of IntList in caffe2 to IntArrayRef
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18282
Differential Revision:
D14569269
Pulled By: bddppq
fbshipit-source-id:
5fc33701b83f9efdec4b456d2691764831d10e7f
Yinghai Lu [Thu, 21 Mar 2019 22:28:20 +0000 (15:28 -0700)]
Blacklist certain op types when doing bound shape inference (#18290)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18290
Some such as `Tile` will mess up our tracking of batch size and for now it makes sense to stop the shape inference on these ops so that we don't lower it and downstream ops without proper batch info.
Reviewed By: zrphercule
Differential Revision:
D14463550
fbshipit-source-id:
2792481efa540f2a7dd310e677c213860c3053ca
Sebastian Messmer [Thu, 21 Mar 2019 21:51:38 +0000 (14:51 -0700)]
Fix use of c10::guts::apply (#18159)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18159
In some instances, the call to forward could clash with std::forward. Fully qualify it to make sure it gets the right one
Reviewed By: ezyang
Differential Revision:
D14512189
fbshipit-source-id:
6242607dbe54fcdb93229c1a4aaee8b84a88caa1
Sebastian Messmer [Thu, 21 Mar 2019 21:51:38 +0000 (14:51 -0700)]
Allow using C10_DECLARE_TENSOR_TYPE and C10_DEFINE_TENSOR_TYPE from any namespace (#18158)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18158
They didn't work when called from other namespaces before because they didn't fully specify the c10 namespace.
Reviewed By: ezyang
Differential Revision:
D14512187
fbshipit-source-id:
a496b89a1bbe2b56137cfae03ab94a60f38d7068
Sebastian Messmer [Thu, 21 Mar 2019 21:51:38 +0000 (14:51 -0700)]
Move schema inference to c10 (#18090)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18090
This schema inference is needed by the c10 operator registration mechanism. Move it to c10.
It is going to be used by diffs stacked on top.
Reviewed By: ezyang
Differential Revision:
D14491454
fbshipit-source-id:
0f8ddcdbd91467c8347d315dd443a1ca8b216481
Sebastian Messmer [Thu, 21 Mar 2019 21:51:38 +0000 (14:51 -0700)]
Allow registering same operator schema multiple times (#18038)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18038
Now that we have named overloads, we can allow registering the same function schema multiple times and just check it's identical.
This is going to be used in custom op registration since they register the schema every time a kernel is registered.
Reviewed By: dzhulgakov
Differential Revision:
D14467494
fbshipit-source-id:
2c26cf72a64b65f120afe05e989302ec42597515
vishwakftw [Thu, 21 Mar 2019 21:18:38 +0000 (14:18 -0700)]
Rename trtrs to triangular_solve (#18213)
Summary:
Changelog:
- Renames `trtrs` to `triangular_solve` to remain consistent with `cholesky_solve` and `solve`.
- Rename all tests, fix callsites
- Create a tentative alias for `triangular_solve` under the name `trtrs`, and add a deprecation warning to not promote usage.
- Move `isnan` to _torch_docs.py
- Remove unnecessary imports
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18213
Differential Revision:
D14566902
Pulled By: ezyang
fbshipit-source-id:
544f57c29477df391bacd5de700bed1add456d3f
kshitij12345 [Thu, 21 Mar 2019 20:10:34 +0000 (13:10 -0700)]
Fix contribution_guide docs (#18237)
Summary:
Fixes Typo and a Link in the `docs/source/community/contribution_guide.rst`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18237
Differential Revision:
D14566907
Pulled By: ezyang
fbshipit-source-id:
3a75797ab6b27d28dd5566d9b189d80395024eaf
svcscm [Thu, 21 Mar 2019 20:08:10 +0000 (13:08 -0700)]
Updating submodules
Reviewed By: yns88
fbshipit-source-id:
80b00c33e6f6c7cfa08f645cd33419f6545f45d2
Xiaomeng Yang [Thu, 21 Mar 2019 19:56:20 +0000 (12:56 -0700)]
Optimize group_norm_op (#17945)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17945
Optimize group_norm_op
Reviewed By: houseroad
Differential Revision:
D14419908
fbshipit-source-id:
4024b5c5dbeff97f4f026d61fc44af1f0e98ed68
Edward Yang [Thu, 21 Mar 2019 19:37:00 +0000 (12:37 -0700)]
Enable running of slow tests in CI. (#18236)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18236
ghimport-source-id:
2bb80d017c2ea833669a2d55b340a922b2d44685
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18236 Enable running of slow tests in CI.**
* #18231 Add a decorator for marking slow tests.
These tests only run on master, as they are slow.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14563115
fbshipit-source-id:
f54ddef4abedc7e872e58657fc9ac537952773d0
Pieter Noordhuis [Thu, 21 Mar 2019 18:49:21 +0000 (11:49 -0700)]
Run clang-format on torch/csrc/distributed/c10d
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18255
Differential Revision:
D14563072
Pulled By: pietern
fbshipit-source-id:
bd83f90ae949b14bc95f4009ba12319c9b7936d0
Edward Yang [Thu, 21 Mar 2019 18:36:12 +0000 (11:36 -0700)]
Shut up compiler about unused the_type. (#18278)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18278
ghimport-source-id:
3c35f6e7229c3c2b3a27d96370d7c05fad58365e
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18278 Shut up compiler about unused this_type.**
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14563050
fbshipit-source-id:
4b516f6c9ef3784d1430f793f304066c351b1a93
Edward Yang [Thu, 21 Mar 2019 18:08:11 +0000 (11:08 -0700)]
Add a decorator for marking slow tests. (#18231)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18231
ghimport-source-id:
78c230f60c41877fe91b89c8c979b160f36f856b
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18231 Add a decorator for marking slow tests.**
The general strategy:
- It's a normal skip decorator, which triggers a skip if
PYTORCH_TEST_WITH_SLOW is not set.
- It also annotates the method in question that says it's
slow. We use this to implement a catch-all skipper in
setUp that skips all non-slow tests when
PYTORCH_TEST_SKIP_FAST is set.
I added a little smoketest to test_torch and showed that I get:
```
Ran 432 tests in 0.017s
OK (skipped=431)
```
when running with PYTORCH_TEST_WITH_SLOW=1 and PYTORCH_TEST_SKIP_FAST=1
CI integration coming in later patch, as well as nontrivial uses of
this decorator.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14544441
fbshipit-source-id:
54435ce4ec827193e019887178c09ebeae3ae2c9
Igor Fedan [Thu, 21 Mar 2019 18:04:15 +0000 (11:04 -0700)]
lint changes
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18276
Differential Revision:
D14563385
Pulled By: ifedan
fbshipit-source-id:
12a51dbdb7b9e96be9fefa21fe298796b1ae6b58
Thomas Viehmann [Thu, 21 Mar 2019 16:59:20 +0000 (09:59 -0700)]
move median to ATen (#17637)
Summary:
This moves median to ATen.
- median with dimension reduces to kthvalue
- median without dimension (aka medianall) is implemented in parallel to kthvalue because we would not want to reshape (copying for non-contiguous) and then copy again in kthvalue. We can sue the helper functions we moved from kthvalue.
- `median_cuda` was accidentally already put into ATen in #17544.
- The quickselect algorthm without indices for CPU in TH is now obsolete and removed.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17637
Differential Revision:
D14346510
Pulled By: ezyang
fbshipit-source-id:
c07ad144efbd6b4194179bb1c02635862521d8cb
Edward Yang [Thu, 21 Mar 2019 16:06:30 +0000 (09:06 -0700)]
Fix B903 lint: save memory for data classes with slots/namedtuple (#18184)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18184
ghimport-source-id:
2ce860b07c58d06dc10cd7e5b97d4ef7c709a50d
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18184 Fix B903 lint: save memory for data classes with slots/namedtuple**
* #18181 Fix B902 lint error: invalid first argument.
* #18178 Fix B006 lint errors: using mutable structure in default argument.
* #18177 Fix lstrip bug revealed by B005 lint
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14530872
fbshipit-source-id:
e26cecab3a8545e7638454c28e654e7b82a3c08a
Edward Yang [Thu, 21 Mar 2019 16:06:30 +0000 (09:06 -0700)]
Fix B902 lint error: invalid first argument. (#18181)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18181
ghimport-source-id:
9c23551584a1a1b0b7ac246367f3a7ae1c50b315
Stack from [ghstack](https://github.com/ezyang/ghstack):
* #18184 Fix B903 lint: save memory for data classes with slots/namedtuple
* **#18181 Fix B902 lint error: invalid first argument.**
* #18178 Fix B006 lint errors: using mutable structure in default argument.
* #18177 Fix lstrip bug revealed by B005 lint
A variety of sins were committed:
- Some code was dead
- Some code was actually a staticmethod
- Some code just named it the wrong way
- Some code was purposely testing the omitted case
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14530876
fbshipit-source-id:
292a371d9a76ddc7bfcfd38b6f0da9165290a58e
Edward Yang [Thu, 21 Mar 2019 16:06:30 +0000 (09:06 -0700)]
Fix B006 lint errors: using mutable structure in default argument. (#18178)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18178
ghimport-source-id:
667ee76b418f505fa64b863e52a603c508dcd1bf
Stack from [ghstack](https://github.com/ezyang/ghstack):
* #18184 Fix B903 lint: save memory for data classes with slots/namedtuple
* #18181 Fix B902 lint error: invalid first argument.
* **#18178 Fix B006 lint errors: using mutable structure in default argument.**
* #18177 Fix lstrip bug revealed by B005 lint
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14530874
fbshipit-source-id:
38f4456a085bfe55f2a96fff53028ebd0d621604
Thomas Viehmann [Thu, 21 Mar 2019 15:02:30 +0000 (08:02 -0700)]
Two amendments for the shape analysis (#18271)
Summary:
Two small refinements to the shape analysis:
- `detach` can set requires grad to false for dimensioned tensors (not sure if I would also need to deal with Complete?).
- add `batch_norm_stats`.
I noticed these while looking at what's going on when trying to code batch norm manually. (Hi wanchaol )
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18271
Differential Revision:
D14561303
Pulled By: ezyang
fbshipit-source-id:
64a6879392e77403c44f2ed82f84b6397754d0ea
Edward Yang [Thu, 21 Mar 2019 14:50:45 +0000 (07:50 -0700)]
Fix lstrip bug revealed by B005 lint (#18177)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18177
ghimport-source-id:
fbbf915b66762fc88bc5b541464e71ba27500958
Stack from [ghstack](https://github.com/ezyang/ghstack):
* #18184 Fix B903 lint: save memory for data classes with slots/namedtuple
* #18181 Fix B902 lint error: invalid first argument.
* #18178 Fix B006 lint errors: using mutable structure in default argument.
* **#18177 Fix lstrip bug revealed by B005 lint**
lstrip() doesn't strip a prefix; it strips all of the characters
in the passed in string. B005 lint revealed this. Replaced with
substring operation.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14530873
fbshipit-source-id:
13b3438fcc3cce13b5110730dc3d0b528a52930f
Igor Fedan [Thu, 21 Mar 2019 07:36:26 +0000 (00:36 -0700)]
Backward function for torch.cdist
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/17173
Differential Revision:
D14111482
Pulled By: ifedan
fbshipit-source-id:
d72cfd53c29d0f8cf5f8ad1148d14f3d5abd938e
Lu Fang [Thu, 21 Mar 2019 05:45:57 +0000 (22:45 -0700)]
Fix ONNX symbolic for argmin and argmax (#18261)
Summary:
Fix the problem introduced in https://github.com/pytorch/pytorch/pull/17103
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18261
Reviewed By: bddppq
Differential Revision:
D14558781
Pulled By: houseroad
fbshipit-source-id:
7bb50072e77d1d7b2a93f4011fa1362f26e9df1c
Xiaomeng Yang [Thu, 21 Mar 2019 01:19:09 +0000 (18:19 -0700)]
Update math::Transpose to support tensor with size > 2G (#17670)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17670
Update math::Transpose to support tensor with size > 2G
i-am-not-moving-c2-to-c10
Differential Revision:
D14313624
fbshipit-source-id:
0b4a85b913972e5a8981f0d40d0c539407b98f30
Jongsoo Park [Thu, 21 Mar 2019 00:02:38 +0000 (17:02 -0700)]
handle dst_bin_width==0 case properly (#18240)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18240
For rare cases when dst_bin_width == 0 we should just put all numbers to an arbitrary bin.
Reviewed By: csummersea
Differential Revision:
D14544685
fbshipit-source-id:
02d04ff8bd1555d6cf7e7eeb1196a4ab3325a9e5
Lu Fang [Wed, 20 Mar 2019 23:27:56 +0000 (16:27 -0700)]
Revert
D14114134: [asr] add fbgemm fp16 (fbfcpacked) support, add global_init_net in predictor_export_meta
Differential Revision:
D14114134
Original commit changeset:
112bb2ceb9d3
fbshipit-source-id:
763262c1b78eed88a653caad5adc27d97feb43aa
Gao, Xiang [Wed, 20 Mar 2019 23:18:49 +0000 (16:18 -0700)]
Cleanup arg{min, max} (#17103)
Summary:
Why do we need this workaround? `PythonArgParser` handles these two cases well.
The discussion started at https://github.com/pytorch/pytorch/pull/6201#issuecomment-
378724406. The conclusion at that time by goldsborough was:
> Because we wanted to allow `dim=None` in Python and route to a different function. Essentially the problem was wanting to wrap the C++ function in Python. AFAIK there is no way of translating `dim=None` behavior into C++? So Richard and I came up with this strategy
Maybe at that time `PythonArgParser` was not powerful enough to handle the routing of two function with same name but different C++ signature.
Will keep an eye on the CI.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17103
Differential Revision:
D14523503
Pulled By: VitalyFedyunin
fbshipit-source-id:
cae3e2678062da2eccd93b51d4050578c7a9ab80
Bharat123Rox [Wed, 20 Mar 2019 23:00:11 +0000 (16:00 -0700)]
Added the exception of ignore_index (#18117)
Summary:
Fix #17801 to add an exception regarding `ignore_index` in the documentation for `torch.nn.CrossEntropyLoss` and `torch.nn.NLLLoss`
If any other files/functions are hit, I'd be glad to incorporate the changes there too! 😊
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18117
Differential Revision:
D14542079
Pulled By: ezyang
fbshipit-source-id:
7b918ac61f441dde7d3d6782d080c500cf2097f1
David Riazati [Wed, 20 Mar 2019 21:48:52 +0000 (14:48 -0700)]
Add .get() for dicts (#18238)
Summary:
Fixes #18232
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18238
Differential Revision:
D14546689
Pulled By: driazati
fbshipit-source-id:
ed021e6f54c891d6c734c8f2345f4e83a3c6c905
Pieter Noordhuis [Wed, 20 Mar 2019 21:30:55 +0000 (14:30 -0700)]
Update nccl submodule to 2.4.2 (#17883)
Summary:
Didn't test this. Let's see what happens.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17883
Differential Revision:
D14547470
Pulled By: pietern
fbshipit-source-id:
c35d232f6bcc5a2dce55da636a0acbea5c2725d8
Pieter Noordhuis [Wed, 20 Mar 2019 21:12:31 +0000 (14:12 -0700)]
Reinstate ncclCommDestroy (#17943)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17943
Together with xw285cornell came up with a solution for static destruction
order fiasco that caused the NCCL context to be destroyed **after**
the CUDA context was already destroyed. In this commit we destroy all
cached NCCL contexts as soon as the last NCCL related Caffe2 operator
instance is destructed, thereby avoiding a dependency on static
variable destruction.
Reviewed By: xw285cornell
Differential Revision:
D14429724
fbshipit-source-id:
fe5ce4b02b1002af8d9f57f6fa089b7a80e316ce
Davide Libenzi [Wed, 20 Mar 2019 20:47:41 +0000 (13:47 -0700)]
Enable autograd to recognize the XLA backend as one providing multiple devices (#17847)
Summary:
…e devices, while not being CUDA/HIP.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17847
Differential Revision:
D14545634
Pulled By: ezyang
fbshipit-source-id:
417181bf2ff4f8978544afe2fb6b042e787854ed
Weiyi Zheng [Wed, 20 Mar 2019 20:45:07 +0000 (13:45 -0700)]
add fbgemm fp16 (fbfcpacked) support, add global_init_net in predictor_export_meta (#17905)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17905
support adding op in global_init_net. because pred_init_net is per thread, and just doesn't cut it.
Reviewed By: jspark1105
Differential Revision:
D14114134
fbshipit-source-id:
112bb2ceb9d3d5e663dd430585567f4eaa2db35f
Zhang Dong [Wed, 20 Mar 2019 19:41:17 +0000 (12:41 -0700)]
fixed typo in shape_analysis.cpp (#18227)
Summary:
cc: VitalyFedyunin
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18227
Differential Revision:
D14541764
Pulled By: VitalyFedyunin
fbshipit-source-id:
9477deb1a99e6581f15a4de4d7631d747f56f3a6
Lu Fang [Wed, 20 Mar 2019 19:03:13 +0000 (12:03 -0700)]
Retain the parameter names in ONNX exporter (#17551)
Summary:
So, we will keep the names of ONNX initializers the same as the names in PyTorch state dict.
Later, we will make this as the default behavior.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17551
Reviewed By: dzhulgakov
Differential Revision:
D14491920
Pulled By: houseroad
fbshipit-source-id:
f355c02e1b90d7ebbebf4be7c0fb6ae208ec795f
Alexandr Morev [Wed, 20 Mar 2019 18:12:55 +0000 (11:12 -0700)]
Fix typo in docstring
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18216
Differential Revision:
D14539824
Pulled By: ezyang
fbshipit-source-id:
490b72951a75f3f8b949a2d692d660a3693ee98a
Vishwak Srinivasan [Wed, 20 Mar 2019 18:06:56 +0000 (11:06 -0700)]
Add batched version of trtrs (#18025)
Summary:
- Remove single batch TH/THC implementations
- Remove `_batch_trtrs_lower` from `multivariate_normal`
- Add tests for batched behavior
- Modify trtrs_backward to accommodate for batched case
- Modify docs
In a future PR, this will be renamed to `triangular_solve`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18025
Differential Revision:
D14523004
Pulled By: ifedan
fbshipit-source-id:
11c6a967d107f969b60e5a5c73ce6bb8099ebbe1
Sacha Refshauge [Wed, 20 Mar 2019 16:16:32 +0000 (09:16 -0700)]
Remove GLOO usage when USE_GLOO is OFF
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18203
Differential Revision:
D14540520
Pulled By: soumith
fbshipit-source-id:
f1c96cc563ed1e913040e3e16b109d3e3030128c
peterjc123 [Wed, 20 Mar 2019 16:16:28 +0000 (09:16 -0700)]
Enable 32 bit CPU build on Windows
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/18176
Differential Revision:
D14539884
Pulled By: ezyang
fbshipit-source-id:
0e4bd9c1ef1830cd9bcc40df36b87534f61def08
peter [Wed, 20 Mar 2019 16:12:40 +0000 (09:12 -0700)]
Correct cmake flags passing (#18217)
Summary:
Fixes #18214.
According to the CMake manual, we should pass the arguments first, and put the directory as the last element. Otherwise, these flags may not be passed correctly.
Reference:
1. https://cmake.org/cmake/help/latest/manual/cmake.1.html#synopsis
2. https://stackoverflow.com/a/
27169347
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18217
Differential Revision:
D14540588
Pulled By: ezyang
fbshipit-source-id:
a027f585dde66c5da7bbbe584fa42c3e56027d59
Gregory Chanan [Wed, 20 Mar 2019 15:39:52 +0000 (08:39 -0700)]
Add python_variable._is_view for debugging. (#18197)
Summary:
I don't know if we actually want to expose this or not, but it's useful for debugging.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18197
Reviewed By: ezyang
Differential Revision:
D14530712
Pulled By: gchanan
fbshipit-source-id:
98fdba9cf113738f0db3a198c49365de536b9919
Johannes M Dieterich [Wed, 20 Mar 2019 14:58:11 +0000 (07:58 -0700)]
Do not apply these explicit unroll pragmas for ROCm. (#18204)
Summary:
Loop analysis indicates that there is a runtime trip count and hence
unrolling cannot take place.
This will silence compile-time warnings we have been observing with recent ROCm releases.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18204
Differential Revision:
D14539875
Pulled By: ezyang
fbshipit-source-id:
a7ea7f2a95603754296b76a6b62a154f56f4ad4d
Edward Yang [Wed, 20 Mar 2019 14:33:51 +0000 (07:33 -0700)]
Copy-edit CONTRIBUTING and update. (#18131)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18131
ghimport-source-id:
473dae70f6c236d317bec77d894310c0aa0376ec
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18131 Copy-edit CONTRIBUTING and update.**
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14505049
fbshipit-source-id:
02aeae33c0049889243c56dd0d761487dac2351e
Ailing Zhang [Wed, 20 Mar 2019 03:02:29 +0000 (20:02 -0700)]
fix cosine_similarity (#18168)
Summary:
fixes #18057 according to colesbury 's suggestion. Thanks!
cc: ezyang
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18168
Differential Revision:
D14520953
Pulled By: ailzhang
fbshipit-source-id:
970e6cfb482d857a81721ec1d0ee4a4df84a0450
Elias Ellison [Wed, 20 Mar 2019 02:38:09 +0000 (19:38 -0700)]
Breakup test misc pt2 (#18191)
Summary:
Further breakup test_misc.h. The remaining tests don't directly map to a jit file so I left them in test_misc.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18191
Differential Revision:
D14533442
Pulled By: eellison
fbshipit-source-id:
7f538ce0aea208b6b55a4716dfcf039548305041
David Riazati [Tue, 19 Mar 2019 23:42:54 +0000 (16:42 -0700)]
Add serialization docs to jit/README (#17951)
Summary:
Documents the serialization format for `torch.jit.save`. Some of the info is copied from houseroad's internal doc.
[Formatted Markdown](https://github.com/driazati/pytorch/blob/serial_docs/torch/csrc/jit/README.md)
Also refactors the readme to have a heading hierarchy + table of contents
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17951
Differential Revision:
D14531644
Pulled By: driazati
fbshipit-source-id:
cbcd9462054cc9f8a2f8cea2c98d8aba4e7d227c
Edward Yang [Tue, 19 Mar 2019 21:47:24 +0000 (14:47 -0700)]
Turn on Travis builds for ghstack PRs. (#18193)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18193
ghimport-source-id:
540859cf0b238a9832f45b3f4c2351e3343fc1a2
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18193 Turn on Travis builds for ghstack PRs.**
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14529945
fbshipit-source-id:
4476e996e311a04f2a997ca9b7c4cf2157dd6286
Michael Suo [Tue, 19 Mar 2019 21:32:52 +0000 (14:32 -0700)]
do not throw when unicode is seen in pull request info (#18195)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18195
ghimport-source-id:
05102cb115c6bd6d141f51905e20155bcd79a908
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18195 [build] do not throw when unicode is seen in pull request info**
Differential Revision:
D14529707
fbshipit-source-id:
2f6a31b01b3a9b044fd24be466cc5325b70929ad
Edward Yang [Tue, 19 Mar 2019 21:15:55 +0000 (14:15 -0700)]
Delete bugbear from Python 2 lint. (#18192)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18192
ghimport-source-id:
9523a09d7ec202ef08cf0ecdf48c42739ea6b0ce
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18192 Delete bugbear from Python 2 lint.**
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Differential Revision:
D14529240
fbshipit-source-id:
1a433b53dd38d1c455e8c0750d97c594ac51ef09
David Riazati [Tue, 19 Mar 2019 20:51:25 +0000 (13:51 -0700)]
Support attributes when emitting function calls (#18156)
Summary:
The type of each `initial_ivalue` is completely known at some point but that information is discarded by the time a call to it is emitted. This PR is kind of a hack, as a better (longer) solution, the method should know about the type of each initial value.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18156
Differential Revision:
D14525768
Pulled By: driazati
fbshipit-source-id:
52d53e9711a07a4551c988bd95fe997e654aa465
Tongzhou Wang [Tue, 19 Mar 2019 20:35:55 +0000 (13:35 -0700)]
Customized pin_memory for PackedSequence (#18079)
Summary:
fixes https://github.com/pytorch/pytorch/issues/18078
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18079
Reviewed By: ezyang
Differential Revision:
D14521192
Pulled By: zou3519
fbshipit-source-id:
cec773a3a6f2c405a0d9701e213b7caf81649181
Edward Yang [Tue, 19 Mar 2019 20:25:04 +0000 (13:25 -0700)]
Enable flake8-bugbear line length checking. (#18138)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18138
ghimport-source-id:
be62a71ef98714e6f168a00f84120f612363528e
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18138 Enable flake8-bugbear line length checking.**
flake8-bugbear's line length checker (B950) which permits violations
of up to 10% but specifies the "true" limit when you go over.
I had to ignore a bunch of flake8-bugbear's other checks when I
turned this on. They're good checks though (they're turned on
in fbcode) and we should fix them eventually.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Reviewed By: salexspb
Differential Revision:
D14508678
fbshipit-source-id:
2610ecc0dd43cc0788d77f4d024ebd85b26b8d41
Michael Suo [Tue, 19 Mar 2019 18:01:05 +0000 (11:01 -0700)]
fix bug in alias analysis (#18146)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18146
ghimport-source-id:
4b061c27c5c44ef0d06066490ed16cab3d0c7a64
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18146 [jit] fix bug in alias analysis**
We handled hasWriters() incorrectly in the case of wildcards. There's
even a comment describing the correct behavior. Sad!
Much thanks to t-vi for tracking this down and suggesting the fix!
Differential Revision:
D14524208
fbshipit-source-id:
8010b54257241bd64013a0d0a8b6e7d22d8c70af
vishwakftw [Tue, 19 Mar 2019 17:36:23 +0000 (10:36 -0700)]
Add backend checks to solve methods (gesv, cholesky_solve) (#18116)
Summary:
Changelog:
- Incorporate a simple backend check in the linearSolveCheckInputs function in LinearAlgebraUtils.h
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18116
Differential Revision:
D14504469
Pulled By: soumith
fbshipit-source-id:
7402b6dbaa8d73048946613b806d54f68bcbd8f4
Hector Yuen [Tue, 19 Mar 2019 17:30:29 +0000 (10:30 -0700)]
fix -Wsign-compare warnings for some files inside c2 (#18123)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18123
the motivation of this fix is to resolve things like:
for(auto i = 0; i < N; i++) where N is bigger than int32
These instances of comparison were found by enabling -Wsign-compare
There are way too many things to fix, so issuing this as a series of fixes
The plan is to fix all these issues and then enable this flag into Caffe2 to catch future instances
Reviewed By: ZolotukhinM
Differential Revision:
D14497094
fbshipit-source-id:
bca3927a2188bd33a508fa503ba221c220cdaefe
Neta Zmora [Tue, 19 Mar 2019 17:29:07 +0000 (10:29 -0700)]
SGD: remove unneeded multiply-add initialization operations (#18114)
Summary:
The momentum buffer is initialized to the value of
d_p, but the current code takes the long way to do this:
1. Create a buffer of zeros
2. Multiply the buffer by the momentum coefficient
3. Add d_p to the buffer
All of these can be collapsed into a single step:
1. Create a clone of d_p
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18114
Differential Revision:
D14509122
Pulled By: ezyang
fbshipit-source-id:
4a79b896201d5ff20770b7ae790c244ba744edb8
Ailing Zhang [Tue, 19 Mar 2019 17:20:06 +0000 (10:20 -0700)]
specialized CUDA impl for dropout in AD (#17756)
Summary:
In aten we have a _fused_dropout implementation for CUDA case. As ngimel suggested if we discard it in JIT AD, it hurts performance.
It doesn't seem ideal to include backend specific implementation in AD, but this is helpful to prevent performance regression atm.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17756
Differential Revision:
D14368999
Pulled By: ailzhang
fbshipit-source-id:
9a371c5020f630e8f6e496849ec9772b6f196169
Neeraj Pradhan [Tue, 19 Mar 2019 17:18:12 +0000 (10:18 -0700)]
Fix underflow issue with dirichlet sample (#17488)
Summary:
Addresses #15738, using fritzo's suggestion. This adds a `torch._sample_dirichlet` method in `Distributions.cpp` and `Distributions.cu`.
- For CPU, this leads to no perf hit since all we do is to promote the `alpha` to double when getting the gamma samples (the gamma sampler anyways uses `accscalar_t`(double for CPU)) and cast it back to float32 on return.
- I have added an analogous method for CUDA as well, but the default sampler for CUDA uses scalar_t for efficiency, so I have kept it as that. With this, I do not see the bias towards 1 as reported in #15738 with `float32`, but there is a spurious mode at 0.5, as would be expected. Users would need to explicitly use `float64` for GPU to not see the spurious mode at 0.5. (EDIT: see note below, it appears that the bias issue is still there for certain builds).
Added some tests and checked that there is no perf regression. My experience with C++ is very limited, so apologies in advance if I missed something basic. cc. ailzhang, fritzo, fmassa
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17488
Differential Revision:
D14410301
Pulled By: ezyang
fbshipit-source-id:
62b2f694b4642685eab06db96d74ce28e05c3992
Gregory Chanan [Tue, 19 Mar 2019 14:57:21 +0000 (07:57 -0700)]
Kill Backend constructor of TensorOptions. (#18137)
Summary:
It's wrong and unused. Use one of the many other constructors instead :).
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18137
Differential Revision:
D14508364
Pulled By: gchanan
fbshipit-source-id:
19c6ff78ad9d9221d0874425edd02b78627c4ca7
Gregory Chanan [Tue, 19 Mar 2019 14:50:31 +0000 (07:50 -0700)]
Remove deviceTypeToBackend, which is underspecified. (#18135)
Summary:
There are multiple backends for a device type, so we just kill this function.
Also, kill an getNonVariableType instance which was also underspecified.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18135
Differential Revision:
D14507474
Pulled By: gchanan
fbshipit-source-id:
fc791a76d4b851b23d09a070725f3838621eb13d
Gregory Chanan [Tue, 19 Mar 2019 14:36:58 +0000 (07:36 -0700)]
Stop generating unimplemented type methods. (#18144)
Summary:
This gets rid of 'aten_sparse' which was used at one time with legacy THS code, but is now only overloaded in native_parse.py.
The way that 'aten_sparse' worked was wonky -- it extended all backends (default [CPU, CUDA]) to include sparse.
But this is totally unnecessary; we already have the backends we need to generate for from type_method_definition_dispatch.
codegen changes: https://github.com/gchanan/pytorch/blob/
fc37c8e171b7ebd1b1755469cf6a146a2abedc13/diff.txt
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18144
Reviewed By: ezyang
Differential Revision:
D14511324
Pulled By: gchanan
fbshipit-source-id:
8bb4ac4cf0985f8756790779a22bc229e18e8e7f
Bharat Raghunathan [Tue, 19 Mar 2019 14:05:29 +0000 (07:05 -0700)]
Corrected type of 'swap' in torch.nn.TripletMarginLoss (#18115)
Summary:
Fix #16428 by correcting type of 'swap' from `float` to `bool`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18115
Differential Revision:
D14516615
Pulled By: ezyang
fbshipit-source-id:
c61a45d533f3a443edf3c31c1ef3d9742bf46d2b
Deepali Chourasia [Tue, 19 Mar 2019 06:06:03 +0000 (23:06 -0700)]
handle scenario when GPU support is not available and p2p_access_pattern is empty (#17974)
Summary:
Observed that when there is no GPU support available `workspace `sets `GetGpuPeerAccessPattern `to `[]` in
https://github.com/pytorch/pytorch/blob/master/caffe2/python/workspace.py#L79
and this case is not handled in https://github.com/pytorch/pytorch/blob/master/caffe2/python/data_parallel_model.py#L1065.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17974
Differential Revision:
D14517066
Pulled By: ezyang
fbshipit-source-id:
186911d95c07e9a55ab82a41d0c7c919e4281bb4
Lutz Roeder [Tue, 19 Mar 2019 03:51:12 +0000 (20:51 -0700)]
Fix Caffe2 operator schemas (#15462) (#13229) (#18109)
Summary:
Maratyszcza harouwu yinghai
This is broken since #13065. `c_str()` returns a pointer that isn't permanent.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18109
Differential Revision:
D14516622
Pulled By: ezyang
fbshipit-source-id:
7113d92eac4f61479c4c7b323cf78cc8aa00b17e
Junji Hashimoto [Tue, 19 Mar 2019 03:44:05 +0000 (20:44 -0700)]
Increase line-width of Declarations.yaml (#18050)
Summary:
There are some line breaks in schema_string of Declarations.yaml.
Is this valid yaml? I am reading yaml-spec.
It seems that the “|” indicator or single/double quote is required to insert line-break.
https://yaml.org/spec/1.2/spec.html
![image](https://user-images.githubusercontent.com/2469618/
54405834-
1e53ac80-471b-11e9-9925-
be13a109eb46.png)
Could you increase line-width of yaml to avoid newline?
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18050
Differential Revision:
D14516694
Pulled By: ezyang
fbshipit-source-id:
1db9f3bf131b54a783d668de973915892603189e
svcscm [Tue, 19 Mar 2019 03:33:38 +0000 (20:33 -0700)]
Updating submodules
Reviewed By: yns88
fbshipit-source-id:
eeeec4229e05916f2c17e525aee5ac4465ef52db
Zhang Dong [Tue, 19 Mar 2019 03:22:20 +0000 (20:22 -0700)]
delete unnecessary file .gitkeep (#18136)
Summary:
delete unnecessary file .gitkeep in /pytorch/tree/master/torch/csrc/nn
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18136
Differential Revision:
D14516584
Pulled By: ezyang
fbshipit-source-id:
a7555693cb3df1c5e37fcd3ca9bb379a2258f2d1
David Riazati [Tue, 19 Mar 2019 01:15:17 +0000 (18:15 -0700)]
Attribute serialization (#17423)
Summary:
Allows serialization/loading of attributes (`IValue`s of any type).
* metadata (attribute name, type) is stored in the `model.json`
* The binary format is a subset of the `pickle` module that supports the operations necessary for `IValue`s
* Attributes are serialized in the order they are defined on a module to a list in a single `attributes` file, with submodule attributes coming first. This order directly matches the order attributes are listed in `model.json`
* This can be inspected in Python with `pickle.load()` or with `pickletools` (PyTorch need not be installed for this to work)
* A class is used to store a tensor's index into the tensor table of the model, so to unpickle the file you have to use a custom Unpickler:
```python
class TensorID(object):
def __setstate__(self, id):
self.id = id
class JitUnpickler(pickle.Unpickler):
def find_class(self, module, name):
if module == '__main__' and name == 'TensorID':
return TensorID
JitUnpickler(open("my_model/attributes.pkl", "rb")).load()
```
* pickle format: https://svn.python.org/projects/python/trunk/Lib/pickletools.py
* It currently does not support/guarantee that anything saved out with `pickle` (i.e. if you edit `attributes` with `pickle` directly) instead of our tools will be imported correctly
Also will fix #17683 and fix #16367
Followup Work:
* document format / choice of pickle: #17951
* create an example
* list specializations
* int size specializations, large binputs
* do a first pass over attributes to output only necessary `BINPUT` ops
* attribute reassignment (e.g `self.my_attribute = new_value`)
* `tensor.save("some_checkpoint.pkl")` support with tensors embedded in Pickle file
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17423
Differential Revision:
D14470965
Pulled By: driazati
fbshipit-source-id:
6a21a9939efdbe59b4bc57fd31d6d630bab5297e
Jongsoo Park [Mon, 18 Mar 2019 23:28:47 +0000 (16:28 -0700)]
fix bug in pool_dnnlowp_op_avx2.cc (#18141)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18141
VLEN should've been 32
Reviewed By: jianyuh
Differential Revision:
D14510780
fbshipit-source-id:
ddf12746e1c69677a268432432ddb088cc210084
svcscm [Mon, 18 Mar 2019 23:16:51 +0000 (16:16 -0700)]
Updating submodules
Reviewed By: yns88
fbshipit-source-id:
ed297c07c681f5f45d3f99edf48680015ca5b138
Vishwak Srinivasan [Mon, 18 Mar 2019 23:01:02 +0000 (16:01 -0700)]
Rename gesv to solve (#18060)
Summary:
Changelog:
- Renames `gesv` to `solve` to remain consistent with `cholesky_solve`.
- Rename all tests, fix callsites
- Create a tentative alias for `solve` under the name `gesv`, and add a deprecated warning to not promote usage.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18060
Differential Revision:
D14503117
Pulled By: zou3519
fbshipit-source-id:
99c16d94e5970a19d7584b5915f051c030d49ff5
James Reed [Mon, 18 Mar 2019 21:06:31 +0000 (14:06 -0700)]
Modify BeamSearch to support CharSourceEncoder (#18107)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18107
Pull Request resolved: https://github.com/pytorch/translate/pull/396
also:
1. fix issues with OptionalType not having a createWithContainedType (PyTorch diff)
2. Delete tests for ONNX full beam search export (nobody is using it and it just makes things harder. Currently ONNX doesn't support `_unwrap_optional`)
Reviewed By: jmp84
Differential Revision:
D14483771
fbshipit-source-id:
0e37ef1cb5a16d03a535eef808b0488b98802128
Narine Kokhlikyan [Mon, 18 Mar 2019 19:21:52 +0000 (12:21 -0700)]
Circular Convolution Function via circular padding (#17240)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17240
Added circular padding in addition to zero padding to Conv1D, Conv2D and Conv3D based on the solution suggested in: https://github.com/pytorch/pytorch/issues/3858
Reviewed By: ezyang
Differential Revision:
D14126416
fbshipit-source-id:
a2f1587503ee0cfff98d5cb0d5b0a600ef8aaeb4
Thomas Viehmann [Mon, 18 Mar 2019 19:07:38 +0000 (12:07 -0700)]
don't include /usr/include when nvcc is in /usr/bin (#18127)
Summary:
...because gcc will have failures with very strange error messages
if you do.
This affects people with Debian/Ubuntu-provided NVCC, the PR should
not change anything for anyone else.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18127
Differential Revision:
D14504386
Pulled By: soumith
fbshipit-source-id:
1aea168723cdc71cdcfffb3193ee116108ae755e
Michael Suo [Mon, 18 Mar 2019 16:56:25 +0000 (09:56 -0700)]
fix double free in test_jit (#18121)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18121
ghimport-source-id:
70c273bfbcb68f7b25cf87f5614c662960864758
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18121 [jit] fix double free in test_jit**
These definitions used to be in anonymous namespace so they weren't exported from the translation unit. #18071 put those in a `test` namespace so I guess they were getting their destructors called twice on exit somehow. Making them static again fixes the problem.
Reviewed By: ezyang
Differential Revision:
D14498349
fbshipit-source-id:
f969781695dcbebdfcfce667fce5b986222a373e
Huitong Qiu [Mon, 18 Mar 2019 15:49:44 +0000 (08:49 -0700)]
Replace resize_dim with set_sizes_and_strides in THTensor_(squeeze) (#18059)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18059
Replace resize_dim() with set_sizes_and_strides() in `THTensor_(squeeze)` in aten/src/TH/generic/THTensor.cpp and `THCTensor_(squeeze)` in aten/src/THC/generic/THCTensor.cpp
Reviewed By: ezyang
Differential Revision:
D14471066
fbshipit-source-id:
1c8c412ff09246c4df6843736e3bf0279bfadea8
Tongzhou Wang [Mon, 18 Mar 2019 04:36:02 +0000 (21:36 -0700)]
update exp. family doc (#18118)
Summary:
sphinx doesn't understand hyphen. it does not merge the two halves together in html.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18118
Differential Revision:
D14498012
Pulled By: mrshenli
fbshipit-source-id:
d6f4cfddc0a8e3a8f91578da43c26ca9c6fff3ce
Gregory Chanan [Sun, 17 Mar 2019 22:37:42 +0000 (15:37 -0700)]
Change one_hot from IndexTensor to Tensor. (#18073)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18073
ghimport-source-id:
f4dadebafa0423c4c5a0e46c15b38129402d830a
Stack:
* #18072 properly device_guard IndexTensor and BoolTensor.
* **#18073 Change one_hot from IndexTensor to Tensor.**
There is no codegen change.
Reviewed By: ezyang
Differential Revision:
D14485248
fbshipit-source-id:
ee2ba8e5dcbbbaf0214a026c8e7ed4e6712becb0
Gregory Chanan [Sun, 17 Mar 2019 22:37:41 +0000 (15:37 -0700)]
properly device_guard IndexTensor and BoolTensor. (#18072)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18072
ghimport-source-id:
9653731602c72f299e095dd50e3afe6bcc8b01d6
Stack:
* **#18072 properly device_guard IndexTensor and BoolTensor.**
* #18073 Change one_hot from IndexTensor to Tensor.
Currently IndexTensor and BoolTensors do not have device_guards applied to them.
This is bad in the case where the only tensor(s) are IndexTensors or BoolTensors, because no device guard is present.
The only case this currently happens is with one_hot which ends up not mattering because of the way the implementation is written. But I wanted to make sure we are covered here.
Reviewed By: ezyang
Differential Revision:
D14485249
fbshipit-source-id:
e57b28086fa1ad2fdd248bb1220e8a2e42da03e1
Michael Suo [Sun, 17 Mar 2019 21:53:41 +0000 (14:53 -0700)]
fix corner case for optional aliasing (#18093)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18093
ghimport-source-id:
021adc52aa7bfe5fff74531c76a8cd28cab30b2a
Stack:
* **#18093 [jit] fix corner case for optional aliasing**
Occasionally the compiler can insert constant Nones to make types line
up. In that case, don't try to make a pointer from the optional type to
None, since we know statically that None won't be mutated or whatever.
Reviewed By: shannonzhu
Differential Revision:
D14493004
fbshipit-source-id:
6564065f39d99ee5af664f3a0fe235892973d9be
Jianyu Huang [Sat, 16 Mar 2019 22:04:02 +0000 (15:04 -0700)]
Typo fix (#18089)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18089
Typo fix for the fully connected layer documentation.
Reviewed By: jspark1105
Differential Revision:
D14488632
fbshipit-source-id:
ca0271ca0250c1d653ed7f250e8588f7b2ce1056
Duc Ngo [Sat, 16 Mar 2019 19:21:55 +0000 (12:21 -0700)]
Caffe2 - Add flag to fails if float point exceptions is detected in operator runs (#18040)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18040
Add flag to fails if float point exceptions is detected in operator runs
Sample exception
Exception [enforce fail at operator.h:837] !std::fetestexcept(FE_DIVBYZERO). Division by zero floating point exception (FE_DIVBYZERO) reported.
Error from operator:
input: "1" input: "0" output: "out" name: "" type: "Div"
Reviewed By: jspark1105
Differential Revision:
D14467731
fbshipit-source-id:
fad030b1d619a5a661ff2114edb947e4562cecdd