Is it possible to generate the dataset that you used ? I would like to run a benchmark for myself and downloading a 1 TB file from Amazon unfortunately is not an option.