Update cnn.R

R/cnn.R

@@ -132,8 +132,8 @@ cnn<-function (X = NULL, Y = NULL, architecture, loss = c("mse", "mae",
   use_custom_dl<-!is.null(train_data_loader)
   if(use_custom_dl){
     sample_batch = train_data_loader$.iter()$.next()
-    X<- as.array(sample_batch$X)
-    Y<- c(as.matrix(sample_batch$Y))
+    X<- as.array(sample_batch[[1]])
+    Y<- c(as.matrix(sample_batch[[2]]))
     batchsize=2L
     if(!xor(is.null(valid_data_loader), validation!=0)) stop("You need to provided an validation data loader if you want to use a validation split with a custom data loader")
   }