Using Kineto profiler from C++ in pytorch

Hi there,

I’d like to know if someone ever tried to use the kineto/builtin profiler from C++ (TorchScript).
Essentially I’d like to see what’s happening when executing a method on a torchscript module (loaded in C++).

We have a model whose runtime keeps on increasing, and I’d like to see if maybe there’s a memory leak or something related.

Thanks !