File size: 1,682 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 |
from easydict import EasyDict
cartpole_a2c_config = dict(
exp_name='cartpole_a2c_seed0',
env=dict(
collector_env_num=8,
evaluator_env_num=5,
n_evaluator_episode=5,
stop_value=195,
),
policy=dict(
cuda=False,
# (bool) whether use on-policy training pipeline(behaviour policy and training policy are the same)
model=dict(
obs_shape=4,
action_shape=2,
encoder_hidden_size_list=[128, 128, 64],
),
learn=dict(
batch_size=40,
learning_rate=0.001,
# (float) loss weight of the entropy regularization, the weight of policy network is set to 1
entropy_weight=0.01,
),
collect=dict(
# (int) collect n_sample data, train model n_iteration times
n_sample=80,
# (float) the trade-off factor lambda to balance 1step td and mc
gae_lambda=0.95,
),
eval=dict(evaluator=dict(eval_freq=50, )),
),
)
cartpole_a2c_config = EasyDict(cartpole_a2c_config)
main_config = cartpole_a2c_config
cartpole_a2c_create_config = dict(
env=dict(
type='cartpole',
import_names=['dizoo.classic_control.cartpole.envs.cartpole_env'],
),
env_manager=dict(type='base'),
policy=dict(type='a2c'),
)
cartpole_a2c_create_config = EasyDict(cartpole_a2c_create_config)
create_config = cartpole_a2c_create_config
if __name__ == "__main__":
# or you can enter `ding -m serial_onpolicy -c cartpole_a2c_config.py -s 0`
from ding.entry import serial_pipeline_onpolicy
serial_pipeline_onpolicy((main_config, create_config), seed=0)
|