hi,I’m constructing a neural network using libtorch, how can I set the default data type to the Double?
I think the proper approach would be to use the AutoDefaultDtypeMode
RAII in libtorch
via:
AutoDefaultDtypeMode dtype_mode(default_dtype);
auto tensor = torch::tensor(1.f); // should be created in the default_dtype
where default_dtype
is c10::ScalarType
.
Thank you very much!I will try it right away,haha