In this project, firstly , I have created ground truth summary of length of 200 words of 1000 input tweet files belonging to 15 disaster events of 400 words. Furthermore, I have evaluated the state-of-the-art Longformer model using the tweet summary pairs and generated the summary for 10 different disaster events. In addition to this, I have also performed fine tuning of the Longformer model using ROUGE- F1 SCORE to evaluate the performance of different disaster datasets.