torch can be installed from CRAN with:
install.packages("torch")You can also install the development version with:
remotes::install_github("mlverse/torch")At the first package load additional software will be installed. Seealso the fullinstallationguide here.
You can create torch tensors from R objects with thetorch_tensor function and convert them back to R objectswithas_array.
library(torch)x<-array(runif(8),dim =c(2,2,2))y<-torch_tensor(x,dtype =torch_float64())y#> torch_tensor#> (1,.,.) =#> 0.6192 0.5800#> 0.2488 0.3681#>#> (2,.,.) =#> 0.0042 0.9206#> 0.4388 0.5664#> [ CPUDoubleType{2,2,2} ]identical(x,as_array(y))#> [1] TRUEIn the following snippet we let torch, using the autograd feature,calculate the derivatives:
x<-torch_tensor(1,requires_grad =TRUE)w<-torch_tensor(2,requires_grad =TRUE)b<-torch_tensor(3,requires_grad =TRUE)y<- w* x+ by$backward()x$grad#> torch_tensor#> 2#> [ CPUFloatType{1} ]w$grad#> torch_tensor#> 1#> [ CPUFloatType{1} ]b$grad#> torch_tensor#> 1#> [ CPUFloatType{1} ]No matter your current skills it’s possible to contribute totorch development. See thecontributingguide for more information.