Lunar Lander hyperparameters about td3-pytorch-bipedalwalker-v2

[nikhilbarhate99]

######### Hyperparameters #########
env_name = "LunarLanderContinuous-v2"
random_seed = 0
log_interval = 10           # print avg reward after interval
solved_reward = 200          # stop training if avg_reward > solved_reward
save_episode = 500         # keep saving after n episodes
max_episodes = 10000        # max num of episodes
max_timesteps = 500         # max timesteps in one episode

gamma = 0.99                # discount for future rewards
batch_size = 100            # num of transitions sampled from replay buffer
lr = 0.001
exploration_noise = 0.1 
polyak = 0.995              # target policy update parameter (1-tau)
policy_noise = 0.2          # target policy smoothing noise std
noise_clip = 0.5
policy_delay = 2            # delayed policy updates parameter

directory = "./"            # save trained models
filename = "TD3_{}_{}".format(env_name, random_seed)
###################################