(no subject)

[impossiblewizardry]

ok my idea is: no more optimization in R. However, keep doing linear regression and random forest in R. Like the really simple baseline models. But actually optimizing parameters of a model, do that in PyTorch or something.

Learning the boundaries of how much I can get away with doing in R is important. Like doing something in R that gets too complex is how I've made myself miserable many times in the past.