In the basic RNN model for solving a binary class$-$sification problem using sequence data of length T composed of a $500-$dimensional vector as an input, instead of using the $500-$dimensional sequence data as it is$,$ it is reduced to $20$ dimensions using an embedded layer and then used as an input for the RNN$.$ If the dimension of the hidden state h$\_($t$)$ is $10$ and the dimension of the immediately connected output layer is $1.$ That is$,$ the size of the output vector is $1)$ the total number of parameters of each layer in this RNN$.($However$,$ bias term is not used in the embedded layer.$)$