RStudio AI Blog: Please allow me to introduce myself: Torch for R

Last January at rstudio::conf, in that distant past when conferences still used to take place at some physical location, my colleague Daniel gave a talk introducing new features and ongoing development in the tensorflow ecosystem. In the Q&A part, he was asked something unexpected: Were we going to build support for PyTorch? He hesitated; that was in fact the plan, and he had already played around with natively implementing torch tensors at a prior time, but he was not completely certain how well "it" would work. "It", that is an implementation which does not bind to Python Torch, meaning, we don't install the PyTorch wheel and import it via reticulate. Instead, we delegate to the underlying C library libtorch for tensor computations and automatic differentiation, while neural network features – layers, activations, optimizers – are implemented directly in R. Removing the intermediary has at least two benefits: For one, the leaner software stack means fewer possible problems in installation and fewer places to look when troubleshooting.