args = parser.parse_args()
if args.resume is None:
args.output = get_output_folder(args.output, args.env)
else:
args.output = args.resume
bullet = ("Bullet" in args.env)
if bullet:
import pybullet
import pybullet_envs
if args.env == "Paint":
from env import CanvasEnv
env = CanvasEnv()
writer.add_image('circle.png', env.target)
elif args.env == "KukaGym":
env = KukaGymEnv(renders=False, isDiscrete=True)
elif args.env == "LTR":
from osim.env import RunEnv
env = RunEnv(visualize=False)
elif args.discrete:
env = gym.make(args.env)
env = env.unwrapped
else:
env = NormalizedEnv(gym.make(args.env))
# input random seed
if args.seed > 0:
np.random.seed(args.seed)
def save_weights(self):
networks = ['actor', 'critic', 'actor_target', 'critic_target']
for name in networks:
network = getattr(self, name)
network.save_weights('ddpg_' + name + '.npz')
def load_weights(self):
networks = ['actor', 'critic', 'actor_target', 'critic_target']
for name in networks:
network = getattr(self, name)
network.load_weights('ddpg_' + name + '.npz')
if __name__ == '__main__':
from env import CanvasEnv
e = CanvasEnv()
agent = nnagent(
e.action_space,
discount_factor=.96,
)
noise_level = 0.5
def r(ep):
global noise_level, e
# agent.render = True
for i in range(ep):
noise_level *= .995
noise_level = max(1e-3, noise_level)
print('ep', i, '/', ep, 'noise_level', noise_level)