We used the training and validation data sets on a multi$-$layer neural network $($NN$)$ with the following architecture: an input layer with $10$ neurons, a hidden layer $1$ with $12$ neurons, a hidden layer $2$ with $6$ neurons, and an output layer with three neurons. The plot below shows the training and validation loss as the network is being trained on different epochs. According to this plot, the trained weights at what epoch should be selected as the final NN and used on the testing set? Round your response to the nearest $10.$if we select the trained weights at epoch $70$ as the final NN$,$ which of the following statements is correct.$$ The NN is properly trained as the because it has the smallest error for the training dataset.$$ The NN is overfit because it fits the training dataset but has poor fit with new dataset.$$ The NN is under$-$fit because it fits the training dataset but has poor fit with new dataset.