| general_cfg: | |
| algo_name: PPO | |
| device: cuda | |
| env_name: CartPole-v1 | |
| eval_eps: 10 | |
| eval_per_episode: 5 | |
| load_checkpoint: false | |
| load_path: Train_CartPole-v1_PPO-KL_20221206-054757 | |
| max_steps: 200 | |
| mode: train | |
| new_step_api: true | |
| render: false | |
| save_fig: true | |
| seed: 1 | |
| show_fig: false | |
| test_eps: 10 | |
| train_eps: 200 | |
| 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 | |
| kl_alpha: 2 | |
| kl_beta: 1.5 | |
| kl_lambda: 0.5 | |
| kl_target: 0.01 | |
| ppo_type: kl | |
| sgd_batch_size: 64 | |
| train_batch_size: 100 | |