|
from easydict import EasyDict |
|
|
|
gym_soccer_pdqn_config = dict( |
|
exp_name='gym_soccer_pdqn_seed0', |
|
env=dict( |
|
collector_env_num=8, |
|
evaluator_env_num=5, |
|
|
|
act_scale=True, |
|
env_id='Soccer-v0', |
|
n_evaluator_episode=5, |
|
stop_value=0.99, |
|
), |
|
policy=dict( |
|
cuda=True, |
|
priority=False, |
|
|
|
priority_IS_weight=False, |
|
discount_factor=0.99, |
|
nstep=1, |
|
model=dict( |
|
obs_shape=10, |
|
action_shape=dict( |
|
action_type_shape=3, |
|
action_args_shape=5, |
|
), |
|
), |
|
learn=dict( |
|
update_per_collect=500, |
|
batch_size=320, |
|
learning_rate_dis=3e-4, |
|
learning_rate_cont=3e-4, |
|
target_theta=0.001, |
|
update_circle=10, |
|
), |
|
|
|
collect=dict( |
|
|
|
n_sample=3200, |
|
|
|
unroll_len=1, |
|
noise_sigma=0.1, |
|
collector=dict(collect_print_freq=1000, ), |
|
), |
|
eval=dict(evaluator=dict(eval_freq=1000, ), ), |
|
|
|
other=dict( |
|
|
|
eps=dict( |
|
|
|
type='exp', |
|
start=1, |
|
end=0.1, |
|
|
|
decay=int(1e5), |
|
), |
|
replay_buffer=dict(replay_buffer_size=int(1e6), ), |
|
), |
|
) |
|
) |
|
|
|
gym_soccer_pdqn_config = EasyDict(gym_soccer_pdqn_config) |
|
main_config = gym_soccer_pdqn_config |
|
|
|
gym_soccer_pdqn_create_config = dict( |
|
env=dict( |
|
type='gym_soccer', |
|
import_names=['dizoo.gym_soccer.envs.gym_soccer_env'], |
|
), |
|
env_manager=dict(type='subprocess'), |
|
policy=dict(type='pdqn'), |
|
) |
|
gym_soccer_pdqn_create_config = EasyDict(gym_soccer_pdqn_create_config) |
|
create_config = gym_soccer_pdqn_create_config |
|
|
|
if __name__ == "__main__": |
|
|
|
from ding.entry import serial_pipeline |
|
serial_pipeline([main_config, create_config], seed=0) |
|
|