The repository implements NLP workflows including retrieval-based and generative chatbots, BERT/XLNet-based text classification models combining convolutional and recurrent architectures, BERT-enhanced named entity recognition with bidirectional LSTMs and conditional random fields, and data augmentation strategies like back-translation and synonym replacement. All models are implemented in Keras.