Fix issues in torch::tensor constructor by yf225 · Pull Request #26890 · pytorch/pytorch

yf225 · 2019-09-26T14:10:50Z

This PR contains the following:

Fix ambiguous overload problem when torch::tensor({{1, 2}}) is used:

../test/cpp/api/tensor.cpp: In member function ‘virtual void TensorTest_MultidimTensorCtor_Test::TestBody()’:
../test/cpp/api/tensor.cpp:202:41: error: call of overloaded ‘tensor(<brace-enclosed initializer list>)’ is ambiguous
     auto tensor = torch::tensor({{1, 2}});
                                         ^
In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0,
                 from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3,
                 from ../test/cpp/api/support.h:7,
                 from ../test/cpp/api/tensor.cpp:2:
../torch/csrc/autograd/generated/variable_factories.h:177:644: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<unsigned char>)
../torch/csrc/autograd/generated/variable_factories.h:177:1603: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<signed char>)
../torch/csrc/autograd/generated/variable_factories.h:177:2562: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<short int>)
../torch/csrc/autograd/generated/variable_factories.h:177:3507: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<int>)
../torch/csrc/autograd/generated/variable_factories.h:177:4450: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<long int>)
../torch/csrc/autograd/generated/variable_factories.h:177:5404: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<float>)
../torch/csrc/autograd/generated/variable_factories.h:177:6354: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<double>)
../torch/csrc/autograd/generated/variable_factories.h:177:7630: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<bool>)
../torch/csrc/autograd/generated/variable_factories.h:177:9224: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::Half>)
../torch/csrc/autograd/generated/variable_factories.h:177:10838: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::BFloat16>)
In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0,
                 from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3,
                 from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3,
                 from ../test/cpp/api/support.h:7,
                 from ../test/cpp/api/tensor.cpp:2:
../torch/csrc/autograd/generated/variable_factories.h:193:19: note: candidate: at::Tensor torch::tensor(torch::detail::InitListTensor)
 inline at::Tensor tensor(detail::InitListTensor list_init_tensor) {
                   ^

After this PR, the multidim tensor constructor torch::tensor(...) should be ready for general use.

This reverts commit 427ea40.

zou3519 · 2019-09-26T20:15:36Z

Fix ambiguous overload problem when torch::tensor({{1, 2}}) is used.

What is the ambiguous overload problem?

zou3519 · 2019-09-26T20:24:19Z

-        scalar_type_(at::k##S),        \
-        type_(InitListTensorType::Scalar) {}
+  // 2. A Tensor represented in `std::vector<ListInitTensor>` form, with value
+  //    `vec()`, Tensor scalar type `scalar_type()`, and Tensor sizes `sizes()`.


It might be better to name vec something more descriptive, like values

zou3519 · 2019-09-26T20:26:49Z

+        sizes_(),                                                                        \
+        scalar_type_(at::k##S),                                                          \
+        type_(ListInitTensorType::Scalar) {}                                             \
+    ListInitTensor(std::vector<T> values) :                                              \


Why do we need this constructor? std::vector<T> is implicitly convertible to at::ArrayRef<T>, so isn't it sufficient to just have that constructor?

I tried removing the constructor that takes in std::vector<T>, but got the following error:

../aten/src/ATen/test/pow_test.cpp:225:39: required from here ../aten/src/ATen/test/pow_test.cpp:111:36: error: no matching function for call to ‘tensor(const std::vector<int>&)’ const auto tensor = torch::tensor(vals); ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../aten/src/ATen/test/pow_test.cpp:5: ../torch/csrc/autograd/generated/variable_factories.h:177:19: note: candidate: at::Tensor torch::tensor(torch::detail::ListInitTensor, const c10::TensorOptions&) inline at::Tensor tensor(detail::ListInitTensor list_init_tensor, const at::TensorOptions& options) { ^ ../torch/csrc/autograd/generated/variable_factories.h:177:19: note: candidate expects 2 arguments, 1 provided ../torch/csrc/autograd/generated/variable_factories.h:181:19: note: candidate: at::Tensor torch::tensor(torch::detail::ListInitTensor) inline at::Tensor tensor(detail::ListInitTensor list_init_tensor) { ^ ../torch/csrc/autograd/generated/variable_factories.h:181:19: note: no known conversion for argument 1 from ‘const std::vector<int>’ to ‘torch::detail::ListInitTensor’

I suspect the issue is that there might be too many conversions going on that the compiler just refuses to do it...

Yikes. Could we make it take const std::vector<T>& values so that we don't incur an additional copy?

In addition, could we just call the ArrayRef ctor here? Something like

ListInitTensor(const std::vector<T>& values) : ListInitTensor(ArrayRef<T>(values)) {}`

This reverts commit 6d4b4ba.

zou3519 · 2019-09-26T21:20:13Z

+        type_(ListInitTensorType::Vector) {                                              \
+      vec_.reserve(values.size());                                                       \
+      for (const auto& elem : values) {                                                  \
+        vec_.push_back(ListInitTensor(elem));                                            \


Discussed offline: there might be a performance regression here. For an ArrayRef with 1000 elements, before we would call at::tensor. Now the code calls tensor.fill_ 1000 times. With 1us dispatch overhead this takes 1(!)ms, which is probably slower than what at::tensor can do (I imagine it does a memcpy).

Oh, interesting. Previously, torch::tensor({1, 2, 3, 4, 5, 6}) went through the ArrayRef ctor. Now, it goes through the ListInitTensor ctor. This means that there might be a performance regression for people who used to use torch::tensor({1, 2, 3, 4, 5, 6})

zou3519

Code looks correct but we can improve it more. I'm not sure how much a perf regression this actually gives (compared to calling at::tensor), it would be good to benchmark and conclude if we should make a fast-path or not.

zou3519 · 2019-09-26T21:29:19Z

-          " with sizes: ",
-          elem.sizes_);
+        type_(ListInitTensorType::Vector) {
+      if (init_list.size() > 0) {


Nit: you can avoid the additional indentation (and reduce the number of lines changed) by doing something like:

if (init_list.size() == 0) { scalar_type_ = c10::ScalarType::Undefined; sizes_ = {0}; return; } // existing code here

This reverts commit c6d29d1.

yf225 · 2019-09-26T23:31:06Z

@zou3519 I reduced the scope of this PR to just add support for torch::tensor({{1, 2}}), because it will require significant refactoring to the InitListTensor code to provide a fast path for 1D tensors. One aspect of the needed refactoring is that we need to change InitListTensor(std::initializer_list<InitListTensor> init_list) to InitListTensor(std::initializer_list<std::initializer_list<InitListTensor>> init_list), otherwise 1D tensors such as torch::tensor({1, 2, 3, 4, ...}) will always take InitListTensor(std::initializer_list<InitListTensor> init_list) over InitListTensor(at::ArrayRef<T> values) and defeats the purpose of the fast path.

This reverts commit cb134ac.

This reverts commit 9b6dd06.

zou3519 · 2019-09-27T17:56:54Z

-      TORCH_CHECK(init_list.size() > 0, "Empty init-list is not supported");
+      TORCH_CHECK(
+        init_list.size() > 0,
+        "Empty init-list is not yet supported. We can create tensors with zero-size dimensions in the following way:\n",


I know I asked you to add this error message, but I'm wondering: what actually happens if a user tries to call torch::tensor({})? Is it a compilation error or does this error message get triggered?

If this error message gets triggered, could we add a test for it?

Compiling torch::tensor({}) gives an "ambiguous overload" compilation error:

../test/cpp/api/tensor.cpp:232:35: error: call of overloaded ‘tensor(<brace-enclosed initializer list>)’ is ambiguous auto tensor = torch::tensor({}); ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:181:644: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<unsigned char>) ../torch/csrc/autograd/generated/variable_factories.h:181:766: note: candidate: at::Tensor torch::tensor(std::initializer_list<unsigned char>) ../torch/csrc/autograd/generated/variable_factories.h:181:887: note: candidate: at::Tensor torch::tensor(uint8_t) ../torch/csrc/autograd/generated/variable_factories.h:181:1603: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<signed char>) ../torch/csrc/autograd/generated/variable_factories.h:181:1724: note: candidate: at::Tensor torch::tensor(std::initializer_list<signed char>) ../torch/csrc/autograd/generated/variable_factories.h:181:1843: note: candidate: at::Tensor torch::tensor(int8_t) ../torch/csrc/autograd/generated/variable_factories.h:181:2562: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<short int>) ../torch/csrc/autograd/generated/variable_factories.h:181:2685: note: candidate: at::Tensor torch::tensor(std::initializer_list<short int>) ../torch/csrc/autograd/generated/variable_factories.h:181:2806: note: candidate: at::Tensor torch::tensor(int16_t) ../torch/csrc/autograd/generated/variable_factories.h:181:3507: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<int>) ../torch/csrc/autograd/generated/variable_factories.h:181:3624: note: candidate: at::Tensor torch::tensor(std::initializer_list<int>) ../torch/csrc/autograd/generated/variable_factories.h:181:3737: note: candidate: at::Tensor torch::tensor(int) ../torch/csrc/autograd/generated/variable_factories.h:181:4450: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<long int>) ../torch/csrc/autograd/generated/variable_factories.h:181:4572: note: candidate: at::Tensor torch::tensor(std::initializer_list<long int>) ../torch/csrc/autograd/generated/variable_factories.h:181:4693: note: candidate: at::Tensor torch::tensor(int64_t) ../torch/csrc/autograd/generated/variable_factories.h:181:5404: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<float>) ../torch/csrc/autograd/generated/variable_factories.h:181:5525: note: candidate: at::Tensor torch::tensor(std::initializer_list<float>) ../torch/csrc/autograd/generated/variable_factories.h:181:5642: note: candidate: at::Tensor torch::tensor(float) ../torch/csrc/autograd/generated/variable_factories.h:181:6354: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<double>) ../torch/csrc/autograd/generated/variable_factories.h:181:6477: note: candidate: at::Tensor torch::tensor(std::initializer_list<double>) ../torch/csrc/autograd/generated/variable_factories.h:181:6596: note: candidate: at::Tensor torch::tensor(double) ../torch/csrc/autograd/generated/variable_factories.h:181:7630: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<bool>) ../torch/csrc/autograd/generated/variable_factories.h:181:7815: note: candidate: at::Tensor torch::tensor(std::initializer_list<bool>) ../torch/csrc/autograd/generated/variable_factories.h:181:8062: note: candidate: at::Tensor torch::tensor(c10::impl::ScalarTypeToCPPType<(c10::ScalarType)11>::type) ../torch/csrc/autograd/generated/variable_factories.h:181:9224: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::Half>) ../torch/csrc/autograd/generated/variable_factories.h:181:9409: note: candidate: at::Tensor torch::tensor(std::initializer_list<c10::Half>) ../torch/csrc/autograd/generated/variable_factories.h:181:9656: note: candidate: at::Tensor torch::tensor(c10::impl::ScalarTypeToCPPType<(c10::ScalarType)5>::type) ../torch/csrc/autograd/generated/variable_factories.h:181:10838: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::BFloat16>) ../torch/csrc/autograd/generated/variable_factories.h:181:11031: note: candidate: at::Tensor torch::tensor(std::initializer_list<c10::BFloat16>) ../torch/csrc/autograd/generated/variable_factories.h:181:11286: note: candidate: at::Tensor torch::tensor(c10::impl::ScalarTypeToCPPType<(c10::ScalarType)15>::type) In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:197:19: note: candidate: at::Tensor torch::tensor(torch::detail::InitListTensor) inline at::Tensor tensor(detail::InitListTensor init_list_tensor) { ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:212:19: note: candidate: at::Tensor torch::tensor(std::initializer_list<torch::detail::InitListTensor>) inline at::Tensor tensor(std::initializer_list<detail::InitListTensor> init_list) { ^

Compiling torch::tensor({{}, {}}) also gives an "ambiguous overload" compilation error:

../test/cpp/api/tensor.cpp:232:41: error: call of overloaded ‘tensor(<brace-enclosed initializer list>)’ is ambiguous auto tensor = torch::tensor({{}, {}}); ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:181:644: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<unsigned char>) ../torch/csrc/autograd/generated/variable_factories.h:181:766: note: candidate: at::Tensor torch::tensor(std::initializer_list<unsigned char>) ../torch/csrc/autograd/generated/variable_factories.h:181:1603: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<signed char>) ../torch/csrc/autograd/generated/variable_factories.h:181:1724: note: candidate: at::Tensor torch::tensor(std::initializer_list<signed char>) ../torch/csrc/autograd/generated/variable_factories.h:181:2562: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<short int>) ../torch/csrc/autograd/generated/variable_factories.h:181:2685: note: candidate: at::Tensor torch::tensor(std::initializer_list<short int>) ../torch/csrc/autograd/generated/variable_factories.h:181:3507: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<int>) ../torch/csrc/autograd/generated/variable_factories.h:181:3624: note: candidate: at::Tensor torch::tensor(std::initializer_list<int>) ../torch/csrc/autograd/generated/variable_factories.h:181:4450: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<long int>) ../torch/csrc/autograd/generated/variable_factories.h:181:4572: note: candidate: at::Tensor torch::tensor(std::initializer_list<long int>) ../torch/csrc/autograd/generated/variable_factories.h:181:5404: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<float>) ../torch/csrc/autograd/generated/variable_factories.h:181:5525: note: candidate: at::Tensor torch::tensor(std::initializer_list<float>) ../torch/csrc/autograd/generated/variable_factories.h:181:6354: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<double>) ../torch/csrc/autograd/generated/variable_factories.h:181:6477: note: candidate: at::Tensor torch::tensor(std::initializer_list<double>) ../torch/csrc/autograd/generated/variable_factories.h:181:7630: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<bool>) ../torch/csrc/autograd/generated/variable_factories.h:181:7815: note: candidate: at::Tensor torch::tensor(std::initializer_list<bool>) ../torch/csrc/autograd/generated/variable_factories.h:181:9224: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::Half>) ../torch/csrc/autograd/generated/variable_factories.h:181:9409: note: candidate: at::Tensor torch::tensor(std::initializer_list<c10::Half>) ../torch/csrc/autograd/generated/variable_factories.h:181:9656: note: candidate: at::Tensor torch::tensor(c10::impl::ScalarTypeToCPPType<(c10::ScalarType)5>::type) ../torch/csrc/autograd/generated/variable_factories.h:181:10838: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::BFloat16>) ../torch/csrc/autograd/generated/variable_factories.h:181:11031: note: candidate: at::Tensor torch::tensor(std::initializer_list<c10::BFloat16>) ../torch/csrc/autograd/generated/variable_factories.h:181:11286: note: candidate: at::Tensor torch::tensor(c10::impl::ScalarTypeToCPPType<(c10::ScalarType)15>::type) In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:197:19: note: candidate: at::Tensor torch::tensor(torch::detail::InitListTensor) inline at::Tensor tensor(detail::InitListTensor init_list_tensor) { ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:212:19: note: candidate: at::Tensor torch::tensor(std::initializer_list<torch::detail::InitListTensor>) inline at::Tensor tensor(std::initializer_list<detail::InitListTensor> init_list) { ^

I think improving the error message here is still a good idea, in case people try to look at the source code to figure out what they should do to create tensors with zero-size dimensions.

Is it clearly documented somewhere that we don't support {}?

Currently it's documented in https://github.com/pytorch/pytorch/pull/26890/files#diff-11228124349c21669d7262f207725b41R184-R188. I will also update https://pytorch.org/cppdocs/notes/tensor_creation.html to document this, after merging this PR.

facebook-github-bot

@yf225 is landing this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

facebook-github-bot · 2019-09-27T19:38:31Z

@yf225 merged this pull request in 2f1932f.

Summary: This PR contains the following: 1. Fix ambiguous overload problem when `torch::tensor({{1, 2}})` is used: ``` ../test/cpp/api/tensor.cpp: In member function ‘virtual void TensorTest_MultidimTensorCtor_Test::TestBody()’: ../test/cpp/api/tensor.cpp:202:41: error: call of overloaded ‘tensor(<brace-enclosed initializer list>)’ is ambiguous auto tensor = torch::tensor({{1, 2}}); ^ In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:177:644: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<unsigned char>) ../torch/csrc/autograd/generated/variable_factories.h:177:1603: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<signed char>) ../torch/csrc/autograd/generated/variable_factories.h:177:2562: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<short int>) ../torch/csrc/autograd/generated/variable_factories.h:177:3507: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<int>) ../torch/csrc/autograd/generated/variable_factories.h:177:4450: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<long int>) ../torch/csrc/autograd/generated/variable_factories.h:177:5404: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<float>) ../torch/csrc/autograd/generated/variable_factories.h:177:6354: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<double>) ../torch/csrc/autograd/generated/variable_factories.h:177:7630: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<bool>) ../torch/csrc/autograd/generated/variable_factories.h:177:9224: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::Half>) ../torch/csrc/autograd/generated/variable_factories.h:177:10838: note: candidate: at::Tensor torch::tensor(c10::ArrayRef<c10::BFloat16>) In file included from ../caffe2/../torch/csrc/api/include/torch/types.h:7:0, from ../caffe2/../torch/csrc/api/include/torch/detail/static.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/pimpl.h:4, from ../caffe2/../torch/csrc/api/include/torch/nn/module.h:3, from ../caffe2/../torch/csrc/api/include/torch/nn/cloneable.h:3, from ../test/cpp/api/support.h:7, from ../test/cpp/api/tensor.cpp:2: ../torch/csrc/autograd/generated/variable_factories.h:193:19: note: candidate: at::Tensor torch::tensor(torch::detail::InitListTensor) inline at::Tensor tensor(detail::InitListTensor list_init_tensor) { ^ ``` After this PR, the multidim tensor constructor `torch::tensor(...)` should be ready for general use. Pull Request resolved: pytorch#26890 Differential Revision: D17632608 Pulled By: yf225 fbshipit-source-id: 2e653d4ad85729d052328a124004d64994bec782

Will Feng added 2 commits September 25, 2019 19:19

add test

715ba71

Fix issues in torch::tensor constructor

9f34a79

yf225 requested a review from zou3519 September 26, 2019 14:10

yf225 requested review from ebetica and goldsborough as code owners September 26, 2019 14:10

pytorchbot added module: cpp Related to C++ API module: internals Related to internal abstractions in c10 and ATen labels Sep 26, 2019

add more test

0e8a210

yf225 added this to the 1.3 milestone Sep 26, 2019

fix type for sizes_

427ea40

yf225 force-pushed the fix_multidim_tensor_ctor branch from bd28b86 to 427ea40 Compare September 26, 2019 18:58

Revert "fix type for sizes_"

73169b6

This reverts commit 427ea40.

yf225 force-pushed the fix_multidim_tensor_ctor branch from 641cde6 to 23401f2 Compare September 26, 2019 19:28

add static_cast

8cb239e

yf225 force-pushed the fix_multidim_tensor_ctor branch from 23401f2 to 8cb239e Compare September 26, 2019 19:41

zou3519 reviewed Sep 26, 2019

View reviewed changes

Comment thread test/cpp/api/tensor.cpp

zou3519 reviewed Sep 26, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h Outdated

zou3519 reviewed Sep 26, 2019

View reviewed changes

Will Feng added 2 commits September 26, 2019 16:51

try to remove ListInitTensor(std::vector<T> values)

6d4b4ba

Revert "try to remove ListInitTensor(std::vector<T> values)"

ef32674

This reverts commit 6d4b4ba.

zou3519 reviewed Sep 26, 2019

View reviewed changes

Will Feng added 3 commits September 26, 2019 18:19

fast path for 1D tensors

d15c463

more test

bd24fcc

simplify

bb53bd4

yf225 force-pushed the fix_multidim_tensor_ctor branch 2 times, most recently from d179b41 to d917f5f Compare September 26, 2019 22:33

yf225 force-pushed the fix_multidim_tensor_ctor branch from 40527fb to d9481ad Compare September 26, 2019 23:22

try a different approach

d9ef7b1

This reverts commit c6d29d1.

yf225 force-pushed the fix_multidim_tensor_ctor branch from d9481ad to d9ef7b1 Compare September 26, 2019 23:23

yf225 changed the title ~~Fix issues and add support for empty tensor in torch::tensor constructor~~ Fix issues in torch::tensor constructor Sep 26, 2019

Will Feng added 2 commits September 26, 2019 23:47

DEBUG

cb134ac

Revert "DEBUG"

ee48a8b

This reverts commit cb134ac.

yf225 commented Sep 27, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h Outdated

update error msg

45c4339

yf225 force-pushed the fix_multidim_tensor_ctor branch from 710927f to 45c4339 Compare September 27, 2019 15:10

zou3519 reviewed Sep 27, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h Outdated

zou3519 reviewed Sep 27, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h Outdated

zou3519 reviewed Sep 27, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h

zou3519 reviewed Sep 27, 2019

View reviewed changes

Comment thread tools/autograd/templates/variable_factories.h

Will Feng added 3 commits September 27, 2019 13:06

improve

4092450

DEBUG

9b6dd06

Revert "DEBUG"

5da4629

This reverts commit 9b6dd06.

zou3519 approved these changes Sep 27, 2019

View reviewed changes

Comment thread test/cpp/api/tensor.cpp

Comment thread tools/autograd/templates/variable_factories.h Outdated

zou3519 reviewed Sep 27, 2019

View reviewed changes

Will Feng added 3 commits September 27, 2019 14:23

better comment

c9fc8d5

more test

67d73bd

new line

0958c7b

facebook-github-bot reviewed Sep 27, 2019

View reviewed changes

facebook-github-bot closed this in 2f1932f Sep 27, 2019

facebook-github-bot added the merged label Sep 27, 2019

mruberry added the Merged label Oct 28, 2020

Conversation

yf225 commented Sep 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zou3519 commented Sep 26, 2019

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zou3519 Sep 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zou3519 left a comment

Choose a reason for hiding this comment

Uh oh!

zou3519 Sep 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yf225 commented Sep 26, 2019

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

facebook-github-bot left a comment

Choose a reason for hiding this comment

Uh oh!

facebook-github-bot commented Sep 27, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

yf225 commented Sep 26, 2019 •

edited

Loading

zou3519 Sep 26, 2019 •

edited

Loading

zou3519 Sep 26, 2019 •

edited

Loading