# save the variables
var_dict = get_variables_as_dict(model_vs)
saver = VariableSaver(var_dict, config.save_dir)
saver.save()
print('=' * 30 + 'result' + '=' * 30)
pprint(best_valid_metrics)
if __name__ == '__main__':
# get config obj
config = ExpConfig()
# parse the arguments
arg_parser = ArgumentParser()
register_config_arguments(config, arg_parser)
arg_parser.parse_args(sys.argv[1:])
config.x_dim = get_data_dim(config.dataset)
print_with_title('Configurations', pformat(config.to_dict()), after='\n')
# open the result object and prepare for result directories if specified
results = MLResults(config.result_dir)
results.save_config(config) # save experiment settings for review
results.make_dirs(config.save_dir, exist_ok=True)
with warnings.catch_warnings():
# suppress DeprecationWarning from NumPy caused by codes in TensorFlow-Probability
warnings.filterwarnings("ignore",
category=DeprecationWarning,
module='numpy')
main()
def main():
config = ExpConfig_transfer()
# parse the arguments
arg_parser = ArgumentParser()
register_config_arguments(config, arg_parser)
arg_parser.parse_args(sys.argv[1:])
executable_action_list = (config.executable_action.replace(" ",
"")).split(',')
dataset = [str(i) for i in range(533)]
if '1' in executable_action_list:
train_part1(
dataset,
start_time=datetime_to_timestamp(config.train_start_time),
last_time=datetime_to_timestamp(config.train_last_time),
action1_model_dir=config.action1_model_dir,
action1_result_dir=config.action1_result_dir,
action1_GPU_device_number=config.action1_GPU_device_number,
action1_sample_ratio=config.action1_sample_ratio,
action1_machine_sample_ratio=config.action1_machine_sample_ratio)
if '2' in executable_action_list:
get_all_machines_z(
dataset,
start_time=datetime_to_timestamp(config.train_z_start_time),
last_time=datetime_to_timestamp(config.train_z_last_time),
action1_model_dir=config.action1_model_dir,
action1_result_dir=config.action1_result_dir,
action2_run_parallel_number=config.action2_run_parallel_number)
if '3' in executable_action_list:
train_part2(
dataset,
action1_result_dir=config.action1_result_dir,
action3_z_file_dir=config.action3_z_file_dir,
action3_z_distance_matrix_name=config.
action3_z_distance_matrix_name,
action3_cluster_number=config.action3_cluster_number,
action3_cluster_png_filename=config.action3_cluster_png_filename,
action3_cluster_result_filename=config.
action3_cluster_result_filename,
action3_machine_file_name=config.action3_machine_file_name)
if '4' in executable_action_list:
train_part3(
dataset,
start_time=datetime_to_timestamp(config.train_start_time),
last_time=datetime_to_timestamp(config.train_last_time),
action3_z_file_dir=config.action3_z_file_dir,
action3_cluster_result_filename=config.
action3_cluster_result_filename,
action3_machine_file_name=config.action3_machine_file_name,
action4_run_parallel_number=config.action4_run_parallel_number,
action4_model_dir_prefix=config.action4_model_dir_prefix,
action4_result_dir_prefix=config.action4_result_dir_prefix,
action1_model_dir=config.action1_model_dir,
action4_cluster_max_machine=config.action4_cluster_max_machine)
if '5' in executable_action_list:
if config.test_start_timestamp is None:
_test_start_timestamp = datetime_to_timestamp(
config.test_start_time)
_test_last_timestamp = datetime_to_timestamp(config.test_last_time)
else:
_test_start_timestamp = config.test_start_timestamp
_test_last_timestamp = config.test_last_timestamp
train_part4(
dataset,
historical_start_time=datetime_to_timestamp(
config.train_z_start_time),
historical_last_time=datetime_to_timestamp(
config.train_z_last_time),
start_time=_test_start_timestamp,
last_time=_test_last_timestamp,
action3_z_file_dir=config.action3_z_file_dir,
action3_cluster_result_filename=config.
action3_cluster_result_filename,
action3_machine_file_name=config.action3_machine_file_name,
action5_run_parallel_number=config.action5_run_parallel_number,
action4_model_dir_prefix=config.action4_model_dir_prefix,
action4_save_path=config.action4_save_path,
action4_result_dir_prefix=config.action4_result_dir_prefix,
get_historical_data_info_flag=config.
action5_get_historical_data_info_flag,
get_threshold_flag=config.action5_get_threshold_flag)