Why I can't use torch::cat or stack to concatenate to tensor in vs?

I use other function in libtorch runs correctly, but when I use cat or stack , it reports debug error, abort has been called.

	auto b = torch::ones({ 3,4 });
	auto c = torch::zeros({ 3,4 });
	auto stack = torch::stack({ b,c }, 1);
	auto cat = torch::cat({ b,c }, 1);
	cout << b << c << cat << stack;


vs2019, libtorch cpu debug

I cannot reproduce the issue and get this output:

# main.cpp
#include <torch/script.h> // One-stop header.

#include <iostream>


int main(int argc, const char* argv[]) {
  auto b = torch::ones({ 3,4 });
  auto c = torch::zeros({ 3,4 });
  auto stack = torch::stack({ b,c }, 1);
  auto cat = torch::cat({ b,c }, 1);
  std::cout << b << c << cat << stack;

  return 0;
}

Output:

./example-app 
 1  1  1  1
 1  1  1  1
 1  1  1  1
[ CPUFloatType{3,4} ] 0  0  0  0
 0  0  0  0
 0  0  0  0
[ CPUFloatType{3,4} ] 1  1  1  1  0  0  0  0
 1  1  1  1  0  0  0  0
 1  1  1  1  0  0  0  0
[ CPUFloatType{3,8} ](1,.,.) = 
  1  1  1  1
  0  0  0  0

(2,.,.) = 
  1  1  1  1
  0  0  0  0

(3,.,.) = 
  1  1  1  1
  0  0  0  0
[ CPUFloatType{3,2,4} ]

Thanks, I found this bug in version:libtorch-win-shared-with-deps-debug-1.12.1+cpu;Now I try torch::cat or stack in version:libtorch-win-shared-with-deps-debug-1.8.2+cpu, it runs well.

1.10.1+CPU runs well also

@ptrblck
It doesn’t work either. Can you share where I can get the old version library?

I also have this question, libtorch CPU version is OK, but CUDA version got this error

libtorch-win-shared-with-deps-1.12.1 cpu failed!

libtorch-win-shared-with-deps-1.13.1 cpu success !