def run(args):
if args.machine == "Mac":
env = UnityEnvironment(file_name='./Reacher.app',seed=1)
else :
env = UnityEnvironment(file_name='./Reacher_Linux_NoVis/Reacher.x86_64',seed=1)
if torch.cuda.is_available():
device = torch.device('cuda')
else :
device = torch.device('cpu')
print("using device", device)
# get the default brain
brain_name = env.brain_names[0]
brain = env.brains[brain_name]
# reset the environment
env_info = env.reset(train_mode=False)[brain_name]
# number of agents
num_agents = len(env_info.agents)
print('Number of agents:', num_agents)
# size of each action
action_size = brain.vector_action_space_size
print('Size of each action:', action_size)
# examine the state space
states = env_info.vector_observations
state_size = states.shape[1]
print('There are {} agents. Each observes a state with length: {}'.format(states.shape[0], state_size))
print('The state for the first agent looks like:', states[0])
#==========================my version=========================
agent = Agent(a_dim=4, s_dim=33, clip_value=1, device=device) # continuous action clip
agent.load("./pretrained/")
eval(env, agent, brain_name)
env.close()
NOISE_C = 1.1
first_ep = 0
with tf.device('/GPU:0'):
env = Environment("data/u20.txt", SEED)
# env = gym.wrappers.Monitor(e.env, 'video/', video_callable=lambda episode_id: True,force = True)
# video = VideoRecorder(env, "video.mp4"
state_shape = env.state_shape
action_len = env.action_shape[0]
action_scale = None
NOISE = 0.6
# np.random.seed(SEED)
agent = Agent(state_shape, action_len, action_scale)
if continued:
agent.load(path)
agent.summary()
for episode in range(first_ep, EPISODES):
state = env.reset()
state = np.reshape(state, state_shape)
score = 0
# print(state)
# done = False
noise = np.random.normal(NOISE, NOISE / 2,
2) / (1 + pow(NOISE_C, episode + 10))
for st in range(MAX_STEPS):
# while not done :
# env.render()
# video.capture_frame()
action = agent.act(state)