| general_cfg: | |
| algo_name: TD3 | |
| collect_traj: false | |
| device: cuda | |
| env_name: gym | |
| eval_per_episode: 50 | |
| load_checkpoint: false | |
| load_model_step: best | |
| load_path: Train_single_BipedalWalkerHardcore-v3_TD3_20230528-151042 | |
| max_episode: 10000 | |
| max_step: 1000 | |
| mode: train | |
| model_save_fre: 70000 | |
| mp_backend: single | |
| n_learners: 1 | |
| n_workers: 1 | |
| online_eval: true | |
| online_eval_episode: 10 | |
| seed: 2023 | |
| share_buffer: true | |
| algo_cfg: | |
| action_type: DPG | |
| actor_layers: | |
| - activation: relu | |
| layer_size: | |
| - 200 | |
| layer_type: linear | |
| - activation: relu | |
| layer_size: | |
| - 200 | |
| layer_type: linear | |
| actor_lr: 0.0001 | |
| batch_size: 256 | |
| buffer_size: 60000 | |
| buffer_type: REPLAY_QUE | |
| critic_layers: | |
| - activation: relu | |
| layer_size: | |
| - 200 | |
| layer_type: linear | |
| - activation: relu | |
| layer_size: | |
| - 200 | |
| layer_type: linear | |
| critic_lr: 0.00012 | |
| expl_noise: 0.25 | |
| explore_steps: 2048 | |
| gamma: 0.99 | |
| noise_clip: 0.5 | |
| policy_freq: 2 | |
| policy_noise: 0.2 | |
| tau: 0.005 | |
| env_cfg: | |
| id: BipedalWalkerHardcore-v3 | |
| ignore_params: | |
| - wrapper | |
| - ignore_params | |
| render_mode: null | |
| wrapper: null | |