Suppose we use MSE$($mean squared error$)$ error function in this RNN$.$ Compute the error of first step $($input$=$h$).$

Seq$2$Sequence Learning Training time:

Use BPTT to train entire RNN network

$($W$\_$hy$,$ W$\_$hh$,$ W$\_$xh$)$ is same in each time step

Character$-$level language model example

Vocabulary: $[$h$,$e$,$l$,$o$]$

Example training sequence:

$$hello$$