You're working on a regression task within a natural language processing (NLP) setting, using a dataset of 100 million labeled examples. After shuffling the data, you split it into training and test sets using a 90/10 ratio. Once your neural network is trained, you evaluate its performance and find that the root-mean-squared error (RMSE) on the training set is twice as high as on the test set . What should you do to improve the model's performance?