| from easydict import EasyDict | |
| import ding.envs.gym_env | |
| cfg = dict( | |
| exp_name='HalfCheetah-v3-SAC', | |
| seed=0, | |
| env=dict( | |
| env_id='HalfCheetah-v3', | |
| collector_env_num=1, | |
| evaluator_env_num=8, | |
| n_evaluator_episode=8, | |
| stop_value=12000, | |
| env_wrapper='mujoco_default', | |
| ), | |
| policy=dict( | |
| cuda=True, | |
| random_collect_size=10000, | |
| model=dict( | |
| obs_shape=17, | |
| action_shape=6, | |
| twin_critic=True, | |
| action_space='reparameterization', | |
| actor_head_hidden_size=256, | |
| critic_head_hidden_size=256, | |
| ), | |
| learn=dict( | |
| update_per_collect=1, | |
| batch_size=256, | |
| learning_rate_q=1e-3, | |
| learning_rate_policy=1e-3, | |
| learning_rate_alpha=3e-4, | |
| ignore_done=True, | |
| target_theta=0.005, | |
| discount_factor=0.99, | |
| alpha=0.2, | |
| reparameterization=True, | |
| auto_alpha=False, | |
| ), | |
| collect=dict( | |
| n_sample=1, | |
| unroll_len=1, | |
| ), | |
| command=dict(), | |
| eval=dict(), | |
| other=dict(replay_buffer=dict(replay_buffer_size=1000000, ), ), | |
| ), | |
| 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 | |