This is a T5-small model finetuned with XSUM and CNN/DailyMail dataset for abstractive summarization. The way both of the datasets are used is by concatenating both of them and took 20% of overall datasets due to limited resources on finetuning the model. The model achieved 0.25 on ROUGE-1, 0.08 on ROUGE-2, and 0,20 on ROUGE-L and ROUGE-LSum. The low scores on the ROUGE metrics might possibly achieved due to the limit put on the hyperparameters, the amount of data used for training and testing, and other circumstances such as the preprocessing of the data and such. A better and resourceful environment might be improve the performance of the model.
