if __name__ == "__main__":
opt = Options().parse()
opt.output = get_output_folder(opt.output, "Paint")
np.random.seed(opt.seed)
torch.manual_seed(opt.seed)
if torch.cuda.is_available(): torch.cuda.manual_seed_all(opt.seed)
random.seed(opt.seed)
torch.backends.cudnn.deterministic = False
torch.backends.cudnn.benchmark = True
from DRL.ddpg import DDPG
from DRL.multi import fastenv
# fenv = fastenv(args.max_step, args.env_batch, writer, args.canvas_color, args.loss_fcn, args.dataset, args.use_multiple_renderers)
# agent = DDPG(args.batch_size, args.env_batch, args.max_step, \
# args.tau, args.discount, args.rmsize, \
# writer, args.resume, args.output, args.loss_fcn, args.renderer, args.use_multiple_renderers)
fenv = fastenv(opt, writer)
agent = DDPG(opt, writer)
evaluate = Evaluator(opt, writer)
print('observation_space', fenv.observation_space, 'action_space',
fenv.action_space)
summary = 'Loss Function - {}\nRenderer - {}\nResuming Model - {}\nbatch_size - {}\nmax_step - {}\nOutput - {}' \
.format(opt.loss_fcn, opt.renderer, opt.resume, opt.batch_size, opt.max_step, opt.output)
writer.add_text('summary', summary, 0)
writer.add_text('Command Line Arguments', str(opt), 0)
train(agent, fenv, evaluate)
parser.add_argument('--env_batch', default=96, type=int, help='concurrent environment number')
parser.add_argument('--tau', default=0.001, type=float, help='moving average for target network')
parser.add_argument('--max_step', default=40, type=int, help='max length for episode')
parser.add_argument('--noise_factor', default=0.01, type=float, help='noise level for parameter space noise')
parser.add_argument('--validate_interval', default=50, type=int, help='how many episodes to perform a validation')
parser.add_argument('--validate_episodes', default=5, type=int, help='how many episode to perform during validation')
parser.add_argument('--train_times', default=2000000, type=int, help='total traintimes')
parser.add_argument('--episode_train_times', default=10, type=int, help='train times for each episode')
parser.add_argument('--resume', default=None, type=str, help='Resuming model path for testing')
parser.add_argument('--output', default='./model', type=str, help='Resuming model path for testing')
parser.add_argument('--debug', dest='debug', action='store_true', help='print some info')
parser.add_argument('--seed', default=1234, type=int, help='random seed')
args = parser.parse_args()
args.output = get_output_folder(args.output, "Paint")
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if torch.cuda.is_available(): torch.cuda.manual_seed_all(args.seed)
random.seed(args.seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
from DRL.ddpg import DDPG
from DRL.multi import fastenv
fenv = fastenv(args.max_step, args.env_batch, writer)
agent = DDPG(args.batch_size, args.env_batch, args.max_step, \
args.tau, args.discount, args.rmsize, \
writer, args.resume, args.output)
evaluate = Evaluator(args, writer)
print('observation_space', fenv.observation_space, 'action_space', fenv.action_space)
train(agent, fenv, evaluate)
help='Resuming model path for testing')
parser.add_argument('--output',
default='./model',
type=str,
help='Resuming model path for testing')
parser.add_argument('--debug',
dest='debug',
action='store_true',
help='print some info')
parser.add_argument('--seed', default=1234, type=int, help='random seed')
args = parser.parse_args()
args.output = get_output_folder(args.output, "Paint")
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if torch.cuda.is_available(): torch.cuda.manual_seed_all(args.seed)
random.seed(args.seed)
torch.backends.cudnn.deterministic = False
torch.backends.cudnn.benchmark = False
from DRL.ddpg import DDPG
from DRL.multi import fastenv
fenv = fastenv(args.state_dim, args.action_dim, args.max_step,
args.env_batch, writer)
agent = DDPG(args.state_dim, args.merged_state_dim, args.action_dim, args.batch_size, args.env_batch, args.max_step, \
args.tau, args.discount, args.rmsize, \
writer, args.resume, args.output)
evaluate = Evaluator(args, writer)
print('observation_space', fenv.observation_space, 'action_space',
fenv.action_space)
train(agent, fenv, evaluate)