Published Paper Details:

Natural language processing uses paraphrasing extensively in applications including question-answering, machine translation, and text-simplification. Transformers-based models have recently produced cutting-edge results on a variety of natural language processing tasks, including paraphrasing. The Microsoft Research Paraphrase Corpus, MS COCO, and Google PAWS datasets, as well as the Quora Question Pairs dataset, were used to train our paraphrase model in this project, which is built on the T5 and Bart transformers. We use a number of benchmark datasets to validate our model. We also conduct a thorough evaluation of our model's performance and investigate the impact of various hyperparameters and training approaches. Our findings suggest that our model can be used for a wide variety of paraphrasing tasks and show the usefulness of transformers-based models for paraphrasing. We are using T5 model in our project for text generation purpose, Parrot model for small scaled paraphrasing and PAWS model for large scale paraphrasing. Furthermore, in addition to all these features, we have incorporated a novel voice-based text input feature, enhancing the accessibility and usability of our paraphrasing system.