general_cfg:
  algo_name: PPO
  device: cpu
  env_name: LunarLander-v2
  eval_eps: 10
  eval_per_episode: 5
  load_checkpoint: false
  load_path: tasks
  max_steps: 1000
  mode: train
  mp_backend: mp
  new_step_api: true
  render: false
  save_fig: true
  seed: 1
  show_fig: false
  test_eps: 20
  train_eps: 600
  wrapper: null
algo_cfg:
  actor_hidden_dim: 256
  actor_lr: 0.0003
  continuous: false
  critic_hidden_dim: 256
  critic_lr: 0.001
  entropy_coef: 0.01
  eps_clip: 0.2
  gamma: 0.99
  k_epochs: 4
  ppo_type: clip
  sgd_batch_size: 32
  train_batch_size: 256