import random domains = ["clipart", "infograph", "painting", "quickdraw", "real", "sketch"] random.seed(42) for domain in domains: with open(domain + '_train.txt', 'r') as f: domain_data = f.readlines() random.shuffle(domain_data) valid_data = domain_data[0:int(len(domain_data)*0.2)] train_data = domain_data[int(len(domain_data)*0.2):] with open(domain + '_train_fold.txt', 'w') as f: f.write("".join(train_data)) with open(domain + '_valid_fold.txt', 'w') as f: f.write("".join(valid_data))