policy_noise_func=policy_noise)
if c.restart_from_trial is not None:
ddpg.load(save_env.get_trial_model_dir())
logger.info("DDPG framework initialized")
# training
# preparations
env = BipedalWalker()
# begin training
episode = Counter()
episode_finished = False
global_step = Counter()
local_step = Counter()
timer = Timer()
while episode < c.max_episodes:
episode.count()
logger.info("Begin episode {} at {}".format(
episode,
dt.now().strftime("%m/%d-%H:%M:%S")))
# environment initialization
env.reset()
# render configuration
if episode.get() % c.profile_int == 0 and global_step.get(
) > c.ddpg_warmup_steps:
render = True
else:
batch_size=c.ppo_update_batch_size,
learning_rate=c.learning_rate)
if c.restart_from_trial is not None:
ppo.load(save_env.get_trial_model_dir())
logger.info("PPO framework initialized")
# training
# preparations
ctx = get_context("spawn")
pool = Pool(processes=c.workers, context=ctx)
pool.enable_global_find(True)
# begin training
episode = Counter(step=c.ppo_update_int)
timer = Timer()
while episode < c.max_episodes:
first_episode = episode.get()
episode.count()
last_episode = episode.get() - 1
logger.info("Begin episode {}-{} at {}".format(
first_episode, last_episode,
dt.now().strftime("%m/%d-%H:%M:%S")))
# begin trials
def run_trial(episode_num):
# TODO: agent_num cannot be pickled ?
env = BipedalMultiCarrier(agent_num=c.agent_num)
# render configuration
batch_size=c.ddpg_update_batch_size,
learning_rate=0.001)
if c.restart_from_trial is not None:
ddpg.load(save_env.get_trial_model_dir())
logger.info("DDPG framework initialized")
# training
# preparations
env = BipedalWalker()
# begin training
episode = Counter()
episode_finished = False
local_step = Counter()
timer = Timer()
while episode < c.max_episodes:
episode.count()
logger.info("Begin episode {} at {}".format(episode, dt.now().strftime("%m/%d-%H:%M:%S")))
# environment initialization
env.reset()
# batch size = 1
total_reward = 0
state, reward = t.tensor(env.reset(), dtype=t.float32, device=c.device), 0
while not episode_finished and local_step.get() <= c.max_steps:
local_step.count()
writer = global_board.writer
logger.info("Directories prepared.")
ppo = create_models()
logger.info("PPO framework initialized")
# training
# preparations
ctx = get_context("spawn")
pool = Pool(processes=c.workers, context=ctx)
pool.enable_global_find(True)
pool.enable_copy_tensors(False)
# begin training
episode = Counter(step=c.ppo_update_int)
timer = Timer()
while episode < c.max_episodes:
first_episode = episode.get()
episode.count()
last_episode = episode.get() - 1
logger.info("Begin episode {}-{} at {}".format(first_episode, last_episode,
dt.now().strftime("%m/%d-%H:%M:%S")))
# begin trials
def run_trial(episode_num):
config = generate_combat_config(c.map_size)
env = magent.GridWorld(config, map_size=c.map_size)
env.reset()
group_handles = env.get_handles()
ddpg = create_models()
logger.info("DDPG framework initialized")
# training
# preparations
config = generate_combat_config(c.map_size)
env = magent.GridWorld(config, map_size=c.map_size)
env.reset()
# begin training
episode = Counter()
episode_finished = False
global_step = Counter()
local_step = Counter()
timer = Timer()
while episode < c.max_episodes:
episode.count()
logger.info("Begin episode {} at {}".format(
episode,
dt.now().strftime("%m/%d-%H:%M:%S")))
# environment initialization
env.reset()
group_handles = env.get_handles()
generate_combat_map(env, c.map_size, c.agent_ratio, group_handles[0],
group_handles[1])
# render configuration