| from easydict import EasyDict | |
| import ding.envs.gym_env | |
| cfg = dict( | |
| exp_name='SpaceInvadersNoFrameskip-v4-C51', | |
| seed=0, | |
| env=dict( | |
| collector_env_num=8, | |
| evaluator_env_num=8, | |
| n_evaluator_episode=8, | |
| stop_value=10000000000, | |
| env_id='SpaceInvadersNoFrameskip-v4', | |
| frame_stack=4, | |
| env_wrapper='atari_default', | |
| ), | |
| policy=dict( | |
| cuda=True, | |
| priority=False, | |
| model=dict( | |
| obs_shape=[4, 84, 84], | |
| action_shape=6, | |
| 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, ), | |
| ), | |
| ), | |
| wandb_logger=dict( | |
| gradient_logger=True, video_logger=True, plot_logger=True, action_logger=True, return_logger=False | |
| ), | |
| ) | |
| cfg = EasyDict(cfg) | |
| env = ding.envs.gym_env.env | |