File size: 2,345 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 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 |
from easydict import EasyDict
agent_num = 5
collector_env_num = 16
evaluator_env_num = 8
main_config = dict(
exp_name='smac_5m6m_wqmix_seed0',
env=dict(
map_name='5m_vs_6m',
difficulty=7,
reward_only_positive=True,
mirror_opponent=False,
agent_num=agent_num,
collector_env_num=collector_env_num,
evaluator_env_num=evaluator_env_num,
stop_value=0.999,
n_evaluator_episode=32,
manager=dict(
shared_memory=False,
reset_timeout=6000,
),
),
policy=dict(
model=dict(
agent_num=agent_num,
obs_shape=72,
global_obs_shape=98,
action_shape=12,
hidden_size_list=[64],
lstm_type='gru',
dueling=False,
),
learn=dict(
update_per_collect=20,
batch_size=32,
learning_rate=0.0005,
clip_value=5,
target_update_theta=0.008,
discount_factor=0.95,
## for OW Optimistically-Weighted
wqmix_ow=True,
alpha=0.5,
## for CW Centrally-Weighted
# wqmix_ow = False,
# alpha = 0.75,
),
collect=dict(
n_episode=32,
unroll_len=10,
env_num=collector_env_num,
),
eval=dict(env_num=evaluator_env_num, evaluator=dict(eval_freq=100, )),
other=dict(
eps=dict(
type='linear',
start=1,
end=0.05,
decay=1000000,
#decay=50000,
),
replay_buffer=dict(
replay_buffer_size=15000,
# (int) The maximum reuse times of each data
max_reuse=1e+9,
max_staleness=1e+9,
),
),
),
)
main_config = EasyDict(main_config)
create_config = dict(
env=dict(
type='smac',
import_names=['dizoo.smac.envs.smac_env'],
),
env_manager=dict(type='subprocess'),
policy=dict(type='wqmix'),
collector=dict(type='episode', get_train_sample=True),
)
create_config = EasyDict(create_config)
if __name__ == '__main__':
from ding.entry import serial_pipeline
serial_pipeline((main_config, create_config), seed=0)
|