total_steps +=1
print(" learning...")
print(touch_count, " touchs in episode ", i_episode)
finish_episode_learning(model, optimizer)
if log_name:
writer.add_scalar(log_name+"/reward",running_reward,total_steps)
writer.add_scalar(log_name+"/touches",len(observed_touches),total_steps)
writer.add_scalar(log_name+"/average_activated_pixels",np.mean(average_activated_pixels),total_steps)
if i_episode % args.log_interval == 0:
print(' Episode {}\tLast length: {:5d}\tAverage length: {:.2f}'.format(i_episode, step, running_reward))
if running_reward > 5000: #env.spec.reward_threshold:
print(" Solved! Running reward is now {} and the last episode runs to {} time steps!".format(running_reward, t))
break
if model_path:
env.mkdir_p(model_path)
torch.save(model.state_dict(), os.path.join(model_path, 'policy.pkl' ))
torch.save(model.state_dict(), os.path.join(model_path, 'cnn_lstm.pkl' ))
print("touched ", touched_episodes, " times")
elif args.mode == "test" or args.mode == "all":
#test
test_labels = []
predicted_labels = []
steps_to_guess = []
correct = 0
total_correct = 0
total = 0
touched_episodes = 0
for i_episode in range(100):
writer.add_scalar(args.log + "/reward", running_reward,
total_steps)
writer.add_scalar(args.log + "/touches", len(observed_touches),
total_steps)
writer.add_scalar(args.log + "/average_activated_pixels",
np.mean(average_activated_pixels), total_steps)
if i_episode % args.log_interval == 0:
print(' Episode {}\tLast length: {:5d}\tAverage length: {:.2f}'.
format(i_episode, step, running_reward))
if running_reward > 5000: #env.spec.reward_threshold:
print(
" Solved! Running reward is now {} and the last episode runs to {} time steps!"
.format(running_reward, t))
break
if args.model_path:
env.mkdir_p(args.model_path)
torch.save(model.state_dict(),
os.path.join(args.model_path, 'policy.pkl'))
torch.save(model.state_dict(),
os.path.join(args.model_path, 'cnn_lstm.pkl'))
elif args.mode == "test" or args.mode == "all":
#test
test_labels = []
predicted_labels = []
steps_to_guess = []
correct = 0
total_correct = 0
total = 0
for i_episode in range(100):