),
daemon=True) for i in range(NUM_ENVIRONMENTS)
]
for p in processes:
p.start()
steps_statistics = Statistics()
total_reward_statistics = Statistics()
collision_statistics = Statistics()
num_nodes_statistics = Statistics()
depth_statistics = Statistics()
stats_statistics = [Statistics() for _ in range(5)]
planner_value_statistics = defaultdict(lambda: Statistics())
value_statistics = defaultdict(lambda: Statistics())
while len(total_reward_statistics) < 100 or total_reward_statistics.stderr(
) > TARGET_SE:
# Read request and process.
(steps, total_reward, collision, num_nodes, depth,
stats) = socket.recv_pyobj()
socket.send_pyobj(0)
if args.task not in ['DriveHard']:
if not collision:
steps_statistics.append(steps)
else:
steps_statistics.append(steps)
total_reward_statistics.append(total_reward)
collision_statistics.append(collision)
if num_nodes is not None:
num_nodes_statistics.append(num_nodes)
if depth is not None:
gen_model.load_state_dict(torch.load(model_path))
steps_statistics = Statistics()
total_reward_statistics = Statistics()
collision_statistics = Statistics()
macro_length_statistics = Statistics()
num_nodes_statistics = Statistics()
depth_statistics = Statistics()
stats_statistics = [Statistics() for _ in range(5)]
planner_value_statistics = defaultdict(lambda: Statistics())
value_statistics = defaultdict(lambda: Statistics())
completed_iterations = 0
while completed_iterations < ITERATIONS or (
TARGET_SE is not None
and total_reward_statistics.stderr() > TARGET_SE):
# Read request and process.
request = socket.recv_pyobj()
instruction = request[0]
instruction_data = request[1:]
if instruction == 'CALL_GENERATOR':
context = instruction_data[0]
state = instruction_data[1]
with torch.no_grad():
(macro_actions) = gen_model.mode(
torch.tensor(instruction_data[0],
dtype=torch.float,
device=device).unsqueeze(0),