File size: 1,949 Bytes
079c32c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
from easydict import EasyDict
pendulum_dqn_config = dict(
exp_name='pendulum_dqn_seed0',
env=dict(
collector_env_num=10,
evaluator_env_num=5,
# (bool) Scale output action into legal range.
act_scale=True,
n_evaluator_episode=5,
stop_value=-250,
continuous=False,
# The path to save the game replay
# replay_path='./pendulum_dqn_seed0/video',
),
policy=dict(
cuda=False,
load_path='pendulum_dqn_seed0/ckpt/ckpt_best.pth.tar', # necessary for eval
model=dict(
obs_shape=3,
action_shape=11, # mean the action shape is 11, 11 discrete actions
encoder_hidden_size_list=[128, 128, 64],
dueling=True,
),
nstep=1,
discount_factor=0.97,
learn=dict(
batch_size=64,
learning_rate=0.001,
),
collect=dict(n_sample=8),
eval=dict(evaluator=dict(eval_freq=40, )),
other=dict(
eps=dict(
type='exp',
start=0.95,
end=0.1,
decay=10000,
),
replay_buffer=dict(replay_buffer_size=20000, ),
),
),
)
pendulum_dqn_config = EasyDict(pendulum_dqn_config)
main_config = pendulum_dqn_config
pendulum_dqn_create_config = dict(
env=dict(
type='pendulum',
import_names=['dizoo.classic_control.pendulum.envs.pendulum_env'],
),
env_manager=dict(type='base'),
policy=dict(type='dqn'),
replay_buffer=dict(type='deque', import_names=['ding.data.buffer.deque_buffer_wrapper']),
)
pendulum_dqn_create_config = EasyDict(pendulum_dqn_create_config)
create_config = pendulum_dqn_create_config
if __name__ == "__main__":
# or you can enter `ding -m serial -c pendulum_dqn_config.py -s 0`
from ding.entry import serial_pipeline
serial_pipeline((main_config, create_config), seed=0)
|