I’m very interested in saving a model trained in c++ to onnx without having to travel through python. It seems like merging script::Module and nn::Module would allow this.
Per torch::jit::trace C++ equivalent of torch.jit.trace · Issue #14811 · pytorch/pytorch · GitHub
We do plan to merge script::Module and nn::Module in a future release, however!
Is there a way that I can track the progress of this merge?