The code to create the dataset is available here. The dataset used in the paper is available on github
2 PAPERS • 2 BENCHMARKS