return None
action[n][0] = obs[i][0]
action[n][1] = 1
action[n][4] = target_id
n = n + 1
return action
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--ip', help='server ip')
parser.add_argument('--port', help='server port', default="11111")
args = parser.parse_args()
env = sc.MultiAgentEnv(args.ip, args.port, speed=30)
env.seed(123)
agent = ClosetAgent(env.action_space)
episodes = 0
while True:
obs = env.reset()
done = False
while not done:
action = agent.act(obs)
obs, reward, done, info = env.step(action)
episodes += 1
env.close()
optimizer.zero_grad()
for prob in model.saved_probs:
loss += criterion(prob, label)
loss.backward()
optimizer.step()
del model.rewards[:]
del model.saved_actions[:]
del model.saved_probs[:]
if __name__ == '__main__':
max_episode_steps = 500
env = sc.MultiAgentEnv(args.ip,
args.port,
speed=30,
max_episode_steps=max_episode_steps)
env.seed(123)
torch.manual_seed(123)
agent = IndependentAgents(env.action_space)
#vis = visdom.Visdom(env='Independent')
episodes = args.eposides
start = time.time()
log_reward_sum = []
log_reward = []
while True:
obs = env.reset()
'''
done = False