Dear all,
I am new to torch (https://torch.mlverse.org/), and more broadly to building neural networks and deep learning models, so I am still learning. I have been an R programmer for a long long long time, and also understand that torch (or more broadly pytorch) mostly associates with python programming (https://pytorch.org/).
I am curious to know why someone would choose to use the R implementation when the Python implementation is so widely used.
What is posit's long-term vision for torch ? is it planned to be maintained for a long time? Is it just mainly about bringing scalable neural network and deep learning workflows to R ecosystem, or is the idea to support workflows that span both R and Python?
I have seen examples in the documentation showing how to load pre-trained model weights into the R implementation, although the architecture still needs to be coded in R exactly the same way:
I would love to hear more about the future directions, especially from the developers if possible.
Please do not hesitate to direct me to the original discussion if this was ever discussed either in this forum or anywhere else.
Best,
Artür