|
from easydict import EasyDict |
|
|
|
enduro_rainbow_config = dict( |
|
exp_name='enduro_rainbow_seed0', |
|
env=dict( |
|
collector_env_num=8, |
|
evaluator_env_num=8, |
|
n_evaluator_episode=8, |
|
stop_value=10000000000, |
|
env_id='EnduroNoFrameskip-v4', |
|
|
|
frame_stack=4 |
|
), |
|
policy=dict( |
|
cuda=True, |
|
priority=True, |
|
model=dict( |
|
obs_shape=[4, 84, 84], |
|
action_shape=9, |
|
encoder_hidden_size_list=[128, 128, 512], |
|
v_min=-10, |
|
v_max=10, |
|
n_atom=51, |
|
), |
|
nstep=3, |
|
discount_factor=0.99, |
|
learn=dict( |
|
update_per_collect=10, |
|
batch_size=32, |
|
learning_rate=0.0001, |
|
target_update_freq=500, |
|
), |
|
collect=dict(n_sample=100, ), |
|
eval=dict(evaluator=dict(eval_freq=4000, )), |
|
other=dict( |
|
eps=dict( |
|
type='exp', |
|
start=1., |
|
end=0.05, |
|
decay=1000000, |
|
), |
|
replay_buffer=dict(replay_buffer_size=400000, ), |
|
), |
|
), |
|
) |
|
enduro_rainbow_config = EasyDict(enduro_rainbow_config) |
|
main_config = enduro_rainbow_config |
|
enduro_rainbow_create_config = dict( |
|
env=dict( |
|
type='atari', |
|
import_names=['dizoo.atari.envs.atari_env'], |
|
), |
|
env_manager=dict(type='subprocess'), |
|
policy=dict(type='rainbow'), |
|
) |
|
enduro_rainbow_create_config = EasyDict(enduro_rainbow_create_config) |
|
create_config = enduro_rainbow_create_config |
|
|
|
if __name__ == '__main__': |
|
|
|
from ding.entry import serial_pipeline |
|
serial_pipeline((main_config, create_config), seed=0) |
|
|
