def main(num_users=5, loc_ep=10, Numb_Glob_Iters=100, lamb=0, learning_rate=0.01, alg='fedprox', weight=True, batch_size=0, dataset="mnist"): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options(num_users, loc_ep, Numb_Glob_Iters, lamb, learning_rate, alg, weight, batch_size, dataset) # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer t = optimizer(options, learner, dataset) t.train()
def main(num_users=5, loc_ep=10, Numb_Glob_Iters=100, lamb=0, learning_rate=0.01, hyper_learning_rate=0.01, alg='fedprox', weight=True, batch_size=0, dataset="mnist"): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) model = MODEL_TYPE + ".py" if (DATA_SET == "cifar100"): learning_rate = 0.001 # parse command line arguments options, learner_model, trainer = read_options( num_users, loc_ep, Numb_Glob_Iters, lamb, learning_rate, hyper_learning_rate, alg, weight, batch_size, dataset, model) # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer t = trainer(options, learner_model, dataset) t.train()
def main(): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options() # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer for i in range(options['times']): # Set seeds random.seed(1 + i) np.random.seed(12 + i) tf.set_random_seed(123 + i) print('......time for runing......', i) t = optimizer(options, learner, dataset) t.train(i) average_data(num_users=options['clients_per_round'], loc_ep1=options['num_epochs'], Numb_Glob_Iters=options['num_rounds'], lamb=options['lamb'], learning_rate=options['learning_rate'], hyper_learning_rate=options['hyper_learning_rate'], algorithms=options['optimizer'], batch_size=options['batch_size'], dataset=options['dataset'], rho=options['rho'], times=options['times'])
def main(): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options() # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer t = optimizer(options, learner, dataset) t.train()
def main(num_users=5, loc_ep=10, alg='fedprox', weight = False): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options(num_users, loc_ep, alg, weight) # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer t = optimizer(options, learner, dataset) t.train()
def main(num_users=5, loc_ep=10, Numb_Glob_Iters=100, lamb=0, learning_rate=0.01, hyper_learning_rate=0.01, alg='fedprox', weight=True, batch_size=0, times=10, rho=0, dataset="mnist"): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options( num_users, loc_ep, Numb_Glob_Iters, lamb, learning_rate, hyper_learning_rate, alg, weight, batch_size, times, rho, dataset) # read data train_path = os.path.join('data', options['dataset'], 'data', 'train') test_path = os.path.join('data', options['dataset'], 'data', 'test') dataset = read_data(train_path, test_path) # call appropriate trainer for i in range(times): # Set seeds random.seed(1 + i) np.random.seed(12 + i) tf.set_random_seed(123 + i) print('......time for runing......', i) t = optimizer(options, learner, dataset) t.train(i) average_data(num_users=num_users, loc_ep1=loc_ep, Numb_Glob_Iters=Numb_Glob_Iters, lamb=lamb, learning_rate=learning_rate, hyper_learning_rate=hyper_learning_rate, algorithms=alg, batch_size=batch_size, dataset=dataset, rho=rho, times=times)
def main(): # suppress tf warnings tf.logging.set_verbosity(tf.logging.WARN) # parse command line arguments options, learner, optimizer = read_options() # read data path = "/".join(os.path.abspath(__file__).split('/')[:-1]) log_path = os.path.join(os.path.abspath('.'), 'out_new', options['dataset']) if not os.path.exists(log_path): os.makedirs(log_path) train_path = os.path.join(path, 'data/train') test_path = os.path.join(path, 'data/test') dataset = read_data(train_path, test_path) # call trainer t = optimizer(options, learner, dataset) t.train()