def runing(config, _log, game_name):
# config 파일로 부터 args 정보를 로드 합니다.
_config = args_sanity_check(config, _log)
args = SN(**config)
args.device = "cuda" if args.use_cuda else "cpu"
env_name = get_env_name(game_name)
# log 기능을 활성화 합니다.
logger = Logger(_log)
unique_token = "{}__{}".format(
args.name,
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
# 텐서보드 기능을 준비 합니다.
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs/{}".format(game_name))
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# 실험을 시작 합니다.
run_sequential(args, logger, env_name)
def my_main(_run, _config, _log):
global mongo_client
import datetime
unique_token = "{}__{}".format(_config["name"], datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
# run the framework
# run(_run, _config, _log, mongo_client, unique_token)
arglist = parse_args()
logger = Logger(_log)
# configure tensorboard logger
unique_token = "{}__{}".format(arglist.exp_name, datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
use_tensorboard = False
if use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))), "results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
logger.setup_sacred(_run)
train(arglist, logger, _config)
# arglist = convert(_config)
#train(arglist)
# force exit
os._exit(0)
def run(_run, _config, _log):
"""
运行，被main函数调用过来
:param _run:
:type _run:
:param _config:
:type _config:
:param _log:
:type _log:
:return:
:rtype:
"""
# 更改一些config中的默认配置，例如cuda，batch等
_config = args_sanity_check(_config, _log)
# 改成Namespace范围的参数
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
#配置日志
logger = Logger(_log)
_log.info("打印实验参数： ")
experiment_params = pprint.pformat(_config,
indent=4,
width=1)
_log.info("\n\n" + experiment_params + "\n")
# 配置Tensorboard Logger ， eg: 'qmix_env=8_adam_td_lambda__2021-04-28_09-40-29'
unique_token = "{}__{}".format(args.name, datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
# 是否使用tensorboard，使用的话，就配置下存储信息
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))), "results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# 默认情况下日志sacred来管理
logger.setup_sacred(_run)
# 运行和训练
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("退出主程序")
print("停止所有线程")
for t in threading.enumerate():
if t.name != "MainThread":
print("Thread {} is alive! Is daemon: {}".format(t.name, t.daemon))
t.join(timeout=1)
print("Thread joined")
print("退出 script")
# 确实退出状态
os._exit(os.EX_OK)
def run(_run, _config, _log):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
set_device = os.getenv('SET_DEVICE')
if args.use_cuda and set_device != '-1':
if set_device is None:
args.device = "cuda"
else:
args.device = f"cuda:{set_device}"
else:
args.device = "cpu"
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# configure tensorboard logger
unique_token = "{}__{}".format(
args.name,
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# sacred is on by default
logger.setup_sacred(_run)
# Run and train
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
print("Stopping all threads")
for t in threading.enumerate():
if t.name != "MainThread":
print("Thread {} is alive! Is daemon: {}".format(t.name, t.daemon))
t.join(timeout=1)
print("Thread joined")
print("Exiting script")
# Making sure framework really exits
os._exit(os.EX_OK)
def run(_run, _config, _log):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
if args.use_cuda:
th.cuda.set_device(args.device_num)
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# configure tensorboard logger
unique_token = "{}__{}".format(
args.name,
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
args.tb_logs = tb_exp_direc
# args.latent_role_direc = os.path.join(tb_exp_direc, "{}").format('latent_role')
logger.setup_tb(tb_exp_direc)
#dump config to the tb directory
with open(os.path.join(tb_exp_direc, "config.yaml"), "w") as f:
yaml.dump(_config, f, default_flow_style=False)
# sacred is on by default
logger.setup_sacred(_run)
# Run and train
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
print("Stopping all threads")
for t in threading.enumerate():
if t.name != "MainThread":
print("Thread {} is alive! Is daemon: {}".format(t.name, t.daemon))
t.join(timeout=1)
print("Thread joined")
print("Exiting script")
# Making sure framework really exits
os._exit(os.EX_OK)
def run(_run, _config, _log):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# configure tensorboard logger
# unique_token = "{}__{}".format(args.name, datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
try:
map_name = _config["env_args"]["map_name"]
except:
map_name = _config["env_args"]["key"]
unique_token = f"{_config['name']}_seed{_config['seed']}_{map_name}_{datetime.datetime.now()}"
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# sacred is on by default
logger.setup_sacred(_run)
# Run and train
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
print("Stopping all threads")
for t in threading.enumerate():
if t.name != "MainThread":
print("Thread {} is alive! Is daemon: {}".format(t.name, t.daemon))
t.join(timeout=1)
print("Thread joined")
print("Exiting script")
def standard_run(_config, _log, game_name):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
# use_cuda=True by default in default.yaml
args.device = "cuda" if args.use_cuda else "cpu"
#args.device = f"cuda:{args.device_num}" if args.use_cuda else "cpu"
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# configure tensorboard logger
unique_token = f"{args.name}__{datetime.datetime.now():%Y-%m-%d_%H-%M-%S}" # e.g. QMIX_2021-08-04_11-09-04
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", f"tb_logs/{game_name}")
tb_exp_direc = os.path.join(tb_logs_direc, f"{unique_token}")
logger.setup_tb(tb_exp_direc)
# Run and train
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
print("Stopping all threads")
for t in threading.enumerate():
if t.name != "MainThread":
print(f"Thread {t.name} is alive! Is daemon: {t.daemon}")
t.join(timeout=1)
print("Thread joined")
print("Exiting script")
# Making sure framework really exits
#os._exit(os.EX_OK) #The os.EX_* values are UNIX only.
os._exit(0) #For Windows
def my_main(_run, _config, _log):
global mongo_client
import datetime
# arglist = parse_args()
# unique_token = "{}__{}".format(arglist.name, datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
# run the framework
# run(_run, _config, _log, mongo_client, unique_token)
logger = Logger(_log)
# configure tensorboard logger
unique_token = "{}__{}".format(
_config["label"],
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
use_tensorboard = False
if use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
logger.setup_sacred(_run)
_log.info("Experiment Parameters:")
import pprint
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# START THE TRAINING PROCESS
runner = Runner(logger)
runner.load(_config)
runner.reset()
# args = vars(arglist)
runner.run(_config)
# runner.run(args)
# train(arglist, logger, _config)
# arglist = convert(_config)
# train(arglist)
# force exit
os._exit(0)
def run(_run, _config, _log, pymongo_client):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
# configure tensorboard logger
unique_token = "{}__{}".format(
args.name,
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# sacred is on by default
logger.setup_sacred(_run)
# Run and train
if args.cross_play and args.evaluate:
run_sequential_cross(args=args, logger=logger)
else:
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
if pymongo_client is not None:
print("Attempting to close mongodb client")
pymongo_client.close()
print("Mongodb client closed")
def run(_run, _config, _log):
# check args sanity
_config = args_sanity_check(_config, _log)
args = SN(**_config)
args.device = "cuda" if args.use_cuda else "cpu"
# setup loggers
logger = Logger(_log)
_log.info("Experiment Parameters:")
experiment_params = pprint.pformat(_config, indent=4, width=1)
_log.info("\n\n" + experiment_params + "\n")
print("\n=================")
auto_describe = 'env_' + args.env_args['map_name'] + '_' + args.name.split(
'_')[0] + '_' # env map algo
if args.name.split('_')[0] == 'smix':
auto_describe = auto_describe + args.mixer + '_'
# average and sub-average
if args.SubAVG_Agent_flag == 0 and args.SubAVG_Mixer_flag == 0:
pass
else:
if args.SubAVG_Agent_flag == 1:
auto_describe = auto_describe + 'average' + str(
args.SubAVG_Agent_K)
if args.SubAVG_Agent_flag_select < 0:
auto_describe = auto_describe + 'neg' + (
"Mean" if args.SubAVG_Agent_name_select_replacement
== 'mean' else "Zero_start2") + '_'
elif args.SubAVG_Agent_flag_select > 0:
auto_describe = auto_describe + 'pos' + (
"Mean" if args.SubAVG_Agent_name_select_replacement
== 'mean' else "Zero_start2") + '_'
if args.SubAVG_Mixer_flag == 1:
auto_describe = auto_describe + 'mix' + str(args.SubAVG_Mixer_K)
if args.SubAVG_Mixer_flag_select < 0:
auto_describe = auto_describe + 'neg' + (
"Mean" if args.SubAVG_Mixer_name_select_replacement
== 'mean' else "Zero_start2") + '_'
elif args.SubAVG_Mixer_flag_select > 0:
auto_describe = auto_describe + 'pos' + (
"Mean" if args.SubAVG_Mixer_name_select_replacement
== 'mean' else "Zero_start2") + '_'
# double
if args.double_q:
auto_describe = auto_describe + 'Double'
else:
auto_describe = auto_describe + 'vanilla'
args.z_auto_describe = auto_describe
print(auto_describe)
print(args.z_describe)
print("using Sub-AVG Agent:", end=' ')
print("Yes K: %d" %
args.SubAVG_Agent_K if args.SubAVG_Agent_flag else "No",
end=' | ')
print("select > or < average:", end=' ')
print("Yes signal: %d" % args.SubAVG_Agent_flag_select
if args.SubAVG_Agent_flag_select else "No",
end=' | ')
print("replace average_select by :", end=' ')
print("mean" if args.SubAVG_Agent_name_select_replacement ==
'mean' else "zero")
print("using Sub-AVG Mixer:", end=' ')
print("Yes K: %d" %
args.SubAVG_Mixer_K if args.SubAVG_Mixer_flag else "No",
end=' | ')
print("mixer: select > or <:", end=' ')
print("Yes signal: %d" % args.SubAVG_Mixer_flag_select
if args.SubAVG_Mixer_flag_select else "No",
end=' | ')
print("replace mixer_select by :", end=' ')
print("mean" if args.SubAVG_Mixer_name_select_replacement ==
'mean' else "zero")
print("using double_q: ", end=' ')
print('Yes' if args.double_q else "No", end=' | ')
# -----------------------------
print("\n==================\n")
# configure tensorboard logger
unique_token = "{}__{}".format(
args.name,
datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S"))
args.unique_token = unique_token
if args.use_tensorboard:
tb_logs_direc = os.path.join(dirname(dirname(abspath(__file__))),
"results", "tb_logs")
tb_exp_direc = os.path.join(tb_logs_direc, "{}").format(unique_token)
logger.setup_tb(tb_exp_direc)
# sacred is on by default
logger.setup_sacred(_run)
# Run and train
run_sequential(args=args, logger=logger)
# Clean up after finishing
print("Exiting Main")
print("Stopping all threads")
for t in threading.enumerate():
if t.name != "MainThread":
print("Thread {} is alive! Is daemon: {}".format(t.name, t.daemon))
t.join(timeout=1)
print("Thread joined")
print("Exiting script")
# Making sure framework really exits
os._exit(os.EX_OK)
