def generate_train_data(self, no_batches, batch_size,
trajectories_data_path, generate_train_data,
gru_dir):
tl.set_log(gru_dir, 'generate_train_data')
train_data = tl.generate_trajectories(self.env, no_batches, batch_size,
trajectories_data_path)
return train_data
def main():
default_server_ip = "127.0.0.1"
args = mgArg.mg_arg(default_server_ip)
tools.set_log(args.logName, args.notverbose)
while True:
secure_communication = tools.SecureCommunication(
args.RHOST, args.RPORT)
secure_communication.secure_connect()
tools.echo_service(secure_communication, args.RHOST, args.logName)
secure_communication.clean_socket()
try:
fsm_object = fsm_process.ProcessFSM(env)
# ***********************************************************************************
# Generating training data *
# ***********************************************************************************
no_batches = 10000
if args.generate_train_data:
train_data = fsm_object.generate_train_data(
no_batches, args.batch_size, trajectories_data_path,
args.generate_train_data, gru_dir)
# ***********************************************************************************
# GRU Network *
# ***********************************************************************************
if args.gru_train or args.gru_test:
tl.set_log(gru_dir, 'train' if args.gru_train else 'test')
gru_net = GRUNet(len(obs), args.gru_size, int(env.action_space.n))
if args.cuda:
gru_net = gru_net.cuda()
if args.gru_train:
logging.info(['No Training Performed!!'])
logging.warning(
'We assume that we already have a pre-trained model @ {}'.
format(gru_net_path))
tl.write_net_readme(gru_net, gru_dir, info={})
if args.gru_test:
test_performance = fsm_object.test_gru(gru_net, gru_net_path,
args.cuda)
# ***********************************************************************************
# Generating BottleNeck training data *
# acc_data = dict with dates as keys and array with jobs as value
# Get passed args
script.collect_args()
# Process args
if script.is_arg('h') or script.is_arg('help'):
print_help()
sys.exit(0)
if script.is_arg('o'):
PATH_OUTPUT = script.get_arg('o')[0]
if script.is_arg('output'):
PATH_OUTPUT = script.get_arg('output')[0]
if script.is_arg('l'):
tools.set_log(script.get_arg('l')[0])
if script.is_arg('logfile'):
tools.set_log(script.get_arg('logfile')[0])
#Get torque accounting files
files = script.get_arg(None)
if files == None:
tools.error('please pass at least one Torque accounting file!')
print_help()
sys.exit(1)
acc_data = torque.parse_accounting(files)
#Create Object User and Job