Libtorch error only in Release build

Hi!
The following program works fine using Debug build. But on Release build it gives the runtime error:

Unhandled exception thrown: read access violation:
_Right was 0x10.

I tried to compile the program with different compiler options (using Microsoft Visual C++ compiler). It works with /MDd option (debug run-time library), and fails with /MD option (release run-time library)

What can be the reason for that?
Here is full program code:

#include <torch/torch.h>

int main() {
    torch::nn::Sequential(torch::nn::Linear(1, 1));
}

Stacktrace:

>	MyChessCpp.exe!std::operator==<char,std::char_traits<char>,std::allocator<char>>(const std::string & _Left, const std::string & _Right)Line 4378	C++
 	[Embedded frame] MyChessCpp.exe!std::equal_to<std::string>::operator()(const std::string &)Line 91	C++
 	[Embedded frame] MyChessCpp.exe!std::_Uhash_compare<std::string,std::hash<std::string>,std::equal_to<std::string>>::operator()(const std::string &)Line 164	C++
 	[Embedded frame] MyChessCpp.exe!std::_Hash<std::_Umap_traits<std::string,unsigned __int64,std::_Uhash_compare<std::string,std::hash<std::string>,std::equal_to<std::string>>,std::allocator<std::pair<std::string const ,unsigned __int64>>,0>>::_Find_last(const std::string &)Line 1677	C++
 	MyChessCpp.exe!std::_Hash<std::_Umap_traits<std::string,unsigned __int64,std::_Uhash_compare<std::string,std::hash<std::string>,std::equal_to<std::string>>,std::allocator<std::pair<std::string const ,unsigned __int64>>,0>>::count<void>(const std::string & _Keyval)Line 1360	C++
 	MyChessCpp.exe!torch::OrderedDict<std::string,std::shared_ptr<torch::nn::Module>>::insert<std::string,std::shared_ptr<torch::nn::Module>>(std::string && key, std::shared_ptr<torch::nn::Module> && value)Line 363	C++
 	MyChessCpp.exe!torch::OrderedDict<std::string,std::shared_ptr<torch::nn::Module>>::insert(std::string key, std::shared_ptr<torch::nn::Module> && value)Line 373	C++
 	MyChessCpp.exe!torch::nn::Module::register_module<torch::nn::Module>(std::string name, std::shared_ptr<torch::nn::Module> module)Line 649	C++
 	MyChessCpp.exe!torch::nn::SequentialImpl::push_back(std::string name, torch::nn::AnyModule any_module)Line 255	C++
 	[Embedded frame] MyChessCpp.exe!torch::nn::SequentialImpl::push_back(std::string)Line 203	C++
 	MyChessCpp.exe!torch::nn::SequentialImpl::push_back<torch::nn::ReLUImpl>(std::string name, const torch::nn::ModuleHolder<torch::nn::ReLUImpl> & module_holder)Line 236	C++
 	[Embedded frame] MyChessCpp.exe!torch::nn::SequentialImpl::push_back(const torch::nn::ModuleHolder<torch::nn::ReLUImpl> &)Line 229	C++
 	[Embedded frame] MyChessCpp.exe!torch::nn::SequentialImpl::{ctor}(torch::nn::ReLU &&)Line 103	C++
 	[Embedded frame] MyChessCpp.exe!torch::nn::ModuleHolder<torch::nn::SequentialImpl>::{ctor}(torch::nn::ReLU &&)Line 64	C++
 	MyChessCpp.exe!torch::nn::Sequential::Sequential<torch::nn::ReLU,void>(torch::nn::ReLU && __param0)Line 20	C++
 	MyChessCpp.exe!main()Line 4	C++
 	[External code]	

@ivankrylatskoe
This might related to how you build the libtorch.
Please create an issue at https://github.com/pytorch/pytorch/issues.
Please provide more info such as what is your operating system, your complier version, your setup, full error logs, etc.
We will have people who is working on build look into this. thx