def run(): args = parse_args() cfg = load_yaml(args) dir_manager = DirManager(cfg) if cfg["TEST"].get("ENABLE", True): cfg["TEST"]["LOAD"] = True # test_all_model(cfg, dir_manager) # test_update(cfg, dir_manager) test_update_real(cfg, dir_manager)
def main(): args = parse_args() create_dirs(args.model_name, [args.checkpoint_dir, args.log_dir]) sess = tf.Session() logger = Logger(args, sess) model = Model(args, logger) reader = Reader(args, sess, logger) if args.action == 'train': trainer = Trainer(sess, model, reader, args, logger) trainer.train() else: predictor = Estimator(sess, model, reader, args, logger) predictor.predict()
def main(): args = parse_args() create_dirs([args.checkpoint_dir, args.log_dir, args.output_dir + args.model_name]) sess = tf.Session() logger = Logger(sess, args) model = Model(args, logger) reader = Reader(args, sess, logger) if args.action == 'train': trainer = Trainer(sess, model, reader, args, logger) trainer.train() elif args.action == 'train_composer': trainer = ComposerTrainer(sess, model, reader, args, logger) trainer.train() elif args.action == 'predict': predictor = Predictor(sess, model, reader, args, logger) predictor.predict() else: raise ValueError('Invalid action argument')
'embedding_size', 'hidden_size', 'word2vec_size', 'nlayers', 'dropout' ]] languages = utils.get_languages(args.languages, args.rare_modes) for i, (lang, rare_mode) in enumerate(languages): print() print('%d. %s (%s)' % (i, lang, rare_mode)) sample_loss = sample_loss_getter(lang, rare_mode, args) xp, yp = bayesian_optimisation(n_iters, sample_loss, bounds, n_pre_samples=n_pre_samples) opt_results += [get_optimal_loss(lang, rare_mode, xp, yp, args)] write_csv( results, '%s/%s__%s__baysian-results.csv' % (args.rfolder, args.model, args.context)) write_csv( opt_results, '%s/%s__%s__opt-results.csv' % (args.rfolder, args.model, args.context)) write_csv( results, '%s/%s__%s__baysian-results-final.csv' % (args.rfolder, args.model, args.context)) if __name__ == '__main__': args = argparser.parse_args(csv_folder='bayes-opt') optimize_languages(args)
from utils import security, argparser, configs from broker import MQTTBroker __version__ = '1.0' __author__ = 'Victor Krook' if __name__ == "__main__": args = argparser.parse_args() if args.version: print( f'Experimental MQTT-broker by {__author__} version: {__version__}') sys.exit(0) if 'setup' in args.setup: configs.setup() if not configs.config_exists(): configs.setup() conf = configs.get_configs() ip = args.ip if args.ip else conf['ip'] port = args.port if args.port else int(conf['port']) broker = MQTTBroker(ip, port, args.verbose, int(conf['max_requests'])) broker.start()
results = [[ 'lang', 'rare_mode', 'avg_len', 'entropy', 'unconditional_entropy', 'test_loss', 'test_acc', 'val_loss', 'val_acc' ]] languages = utils.get_languages(args.languages, args.rare_modes) for i, (lang, rare_mode) in enumerate(languages): print() print('%d. Language %s (%s)' % (i, lang, rare_mode)) instance_ids = get_ids(lang, rare_mode) avg_len, entropy, uncond_entropy, test_loss, test_acc, \ val_loss, val_acc = run_language_enveloper_cv(lang, rare_mode, instance_ids, args) results += [[ lang, rare_mode, avg_len, entropy, uncond_entropy, test_loss, test_acc, val_loss, val_acc ]] write_csv( results, '%s/%s__%s__results.csv' % (args.rfolder, args.model, args.context)) write_csv( results, '%s/%s__%s__results-final.csv' % (args.rfolder, args.model, args.context)) if __name__ == '__main__': args = argparser.parse_args(csv_folder='cv') run_languages(args)
def run_languages(args): results = [[ 'lang', 'rare_mode', 'avg_len', 'entropy', 'unconditional_entropy', 'test_loss', 'test_acc', 'best_epoch', 'val_loss', 'val_acc' ]] languages = utils.get_languages(args.languages, args.rare_modes) for i, (lang, rare_mode) in enumerate(languages): print() print(i, end=' ') avg_len, entropy, uncond_entropy, test_loss, test_acc, \ best_epoch, val_loss, val_acc = run_language_enveloper(lang, rare_mode, args) results += [[ lang, rare_mode, avg_len, entropy, uncond_entropy, test_loss, test_acc, best_epoch, val_loss, val_acc ]] write_csv( results, '%s/%s__%s__results.csv' % (args.rfolder, args.model, args.context)) write_csv( results, '%s/%s__%s__results-final.csv' % (args.rfolder, args.model, args.context)) if __name__ == '__main__': args = argparser.parse_args(csv_folder='normal') run_languages(args)