Any chance this will be included in an official release of PyTorch?
Con-Mi opened this issue · comments
you never know deeplearnjs was also separate project before it became part of Google and got rebranded as tensorflowjs. The questions here is: Is the beginning right? Are the data structures and architecture right for this?
Thanks for the interest. I mainly aimed to create a light-weight way to run inference on PT models in node. This could be useful for serving the trained model, e.g., writing the application server in JS is much more light-weight than in C++. The training is out of scope for this project at the moment.