2023-01-14 22:25:23 - r - INFO: - Hyperparameters: 2023-01-14 22:25:23 - r - INFO: - ================================================================================ 2023-01-14 22:25:23 - r - INFO: - Name Value Type 2023-01-14 22:25:23 - r - INFO: - env_name CartPole-v1 2023-01-14 22:25:23 - r - INFO: - new_step_api 1 2023-01-14 22:25:23 - r - INFO: - wrapper None 2023-01-14 22:25:23 - r - INFO: - render 0 2023-01-14 22:25:23 - r - INFO: - algo_name C51 2023-01-14 22:25:23 - r - INFO: - mode train 2023-01-14 22:25:23 - r - INFO: - seed 1 2023-01-14 22:25:23 - r - INFO: - device cuda 2023-01-14 22:25:23 - r - INFO: - train_eps 100 2023-01-14 22:25:23 - r - INFO: - test_eps 10 2023-01-14 22:25:23 - r - INFO: - eval_eps 10 2023-01-14 22:25:23 - r - INFO: - eval_per_episode 5 2023-01-14 22:25:23 - r - INFO: - max_steps 200 2023-01-14 22:25:23 - r - INFO: - load_checkpoint 0 2023-01-14 22:25:23 - r - INFO: - load_path Train_CartPole-v1_C51_20221026-054757 2023-01-14 22:25:23 - r - INFO: - show_fig 0 2023-01-14 22:25:23 - r - INFO: - save_fig 1 2023-01-14 22:25:23 - r - INFO: - Vmin 0.0 2023-01-14 22:25:23 - r - INFO: - Vmax 200.0 2023-01-14 22:25:23 - r - INFO: - n_atoms 51 2023-01-14 22:25:23 - r - INFO: - gamma 0.95 2023-01-14 22:25:23 - r - INFO: - tau 1.0 2023-01-14 22:25:23 - r - INFO: - num_atoms 51 2023-01-14 22:25:23 - r - INFO: - support tensor(0.) 2023-01-14 22:25:23 - r - INFO: - delta_z 4.0 2023-01-14 22:25:23 - r - INFO: - batch_size 64 2023-01-14 22:25:23 - r - INFO: - lr 0.0001 2023-01-14 22:25:23 - r - INFO: - target_update 4 2023-01-14 22:25:23 - r - INFO: - memory_capacity 10000 2023-01-14 22:25:23 - r - INFO: - epsilon_start 0.95 2023-01-14 22:25:23 - r - INFO: - epsilon_end 0.01 2023-01-14 22:25:23 - r - INFO: - epsilon_decay 500 2023-01-14 22:25:23 - r - INFO: - value_layers [{'layer_type': 'linear', 'layer_dim': ['n_states', 256], 'activation': 'relu'}, {'layer_type': 'linear', 'layer_dim': [256, 256], 'activation': 'relu'}, {'layer_type': 'linear', 'layer_dim': [256, 'n_actions'], 'activation': 'none'}] 2023-01-14 22:25:23 - r - INFO: - buffer_size 100000 2023-01-14 22:25:23 - r - INFO: - task_dir /home/dingli/rl-tutorials/joyrl/tasks/Train_CartPole-v1_C51_20230114-222523 2023-01-14 22:25:23 - r - INFO: - model_dir /home/dingli/rl-tutorials/joyrl/tasks/Train_CartPole-v1_C51_20230114-222523/models 2023-01-14 22:25:23 - r - INFO: - res_dir /home/dingli/rl-tutorials/joyrl/tasks/Train_CartPole-v1_C51_20230114-222523/results 2023-01-14 22:25:23 - r - INFO: - log_dir /home/dingli/rl-tutorials/joyrl/tasks/Train_CartPole-v1_C51_20230114-222523/logs 2023-01-14 22:25:23 - r - INFO: - traj_dir /home/dingli/rl-tutorials/joyrl/tasks/Train_CartPole-v1_C51_20230114-222523/traj 2023-01-14 22:25:23 - r - INFO: - ================================================================================ 2023-01-14 22:25:23 - r - INFO: - n_states: 4, n_actions: 2 2023-01-14 22:25:25 - r - INFO: - Start training! 2023-01-14 22:25:25 - r - INFO: - Env: CartPole-v1, Algorithm: C51, Device: cuda 2023-01-14 22:25:25 - r - INFO: - Episode: 1/100, Reward: 18.000, Step: 18 2023-01-14 22:25:25 - r - INFO: - Episode: 2/100, Reward: 35.000, Step: 35 2023-01-14 22:25:25 - r - INFO: - Episode: 3/100, Reward: 12.000, Step: 12 2023-01-14 22:25:25 - r - INFO: - Episode: 4/100, Reward: 13.000, Step: 13 2023-01-14 22:25:25 - r - INFO: - Episode: 5/100, Reward: 26.000, Step: 26 2023-01-14 22:25:25 - r - INFO: - Current episode 5 has the best eval reward: 8.800 2023-01-14 22:25:25 - r - INFO: - Episode: 6/100, Reward: 10.000, Step: 10 2023-01-14 22:25:25 - r - INFO: - Episode: 7/100, Reward: 9.000, Step: 9 2023-01-14 22:25:25 - r - INFO: - Episode: 8/100, Reward: 10.000, Step: 10 2023-01-14 22:25:25 - r - INFO: - Episode: 9/100, Reward: 23.000, Step: 23 2023-01-14 22:25:26 - r - INFO: - Episode: 10/100, Reward: 36.000, Step: 36 2023-01-14 22:25:26 - r - INFO: - Current episode 10 has the best eval reward: 10.400 2023-01-14 22:25:26 - r - INFO: - Episode: 11/100, Reward: 16.000, Step: 16 2023-01-14 22:25:26 - r - INFO: - Episode: 12/100, Reward: 14.000, Step: 14 2023-01-14 22:25:26 - r - INFO: - Episode: 13/100, Reward: 14.000, Step: 14 2023-01-14 22:25:26 - r - INFO: - Episode: 14/100, Reward: 13.000, Step: 13 2023-01-14 22:25:26 - r - INFO: - Episode: 15/100, Reward: 21.000, Step: 21 2023-01-14 22:25:26 - r - INFO: - Episode: 16/100, Reward: 14.000, Step: 14 2023-01-14 22:25:26 - r - INFO: - Episode: 17/100, Reward: 8.000, Step: 8 2023-01-14 22:25:26 - r - INFO: - Episode: 18/100, Reward: 29.000, Step: 29 2023-01-14 22:25:26 - r - INFO: - Episode: 19/100, Reward: 22.000, Step: 22 2023-01-14 22:25:26 - r - INFO: - Episode: 20/100, Reward: 22.000, Step: 22 2023-01-14 22:25:27 - r - INFO: - Current episode 20 has the best eval reward: 30.700 2023-01-14 22:25:27 - r - INFO: - Episode: 21/100, Reward: 22.000, Step: 22 2023-01-14 22:25:27 - r - INFO: - Episode: 22/100, Reward: 42.000, Step: 42 2023-01-14 22:25:27 - r - INFO: - Episode: 23/100, Reward: 64.000, Step: 64 2023-01-14 22:25:27 - r - INFO: - Episode: 24/100, Reward: 53.000, Step: 53 2023-01-14 22:25:28 - r - INFO: - Episode: 25/100, Reward: 55.000, Step: 55 2023-01-14 22:25:28 - r - INFO: - Current episode 25 has the best eval reward: 110.200 2023-01-14 22:25:28 - r - INFO: - Episode: 26/100, Reward: 61.000, Step: 61 2023-01-14 22:25:29 - r - INFO: - Episode: 27/100, Reward: 96.000, Step: 96 2023-01-14 22:25:29 - r - INFO: - Episode: 28/100, Reward: 55.000, Step: 55 2023-01-14 22:25:29 - r - INFO: - Episode: 29/100, Reward: 49.000, Step: 49 2023-01-14 22:25:29 - r - INFO: - Episode: 30/100, Reward: 83.000, Step: 83 2023-01-14 22:25:30 - r - INFO: - Episode: 31/100, Reward: 67.000, Step: 67 2023-01-14 22:25:31 - r - INFO: - Episode: 32/100, Reward: 140.000, Step: 140 2023-01-14 22:25:31 - r - INFO: - Episode: 33/100, Reward: 65.000, Step: 65 2023-01-14 22:25:31 - r - INFO: - Episode: 34/100, Reward: 55.000, Step: 55 2023-01-14 22:25:31 - r - INFO: - Episode: 35/100, Reward: 63.000, Step: 63 2023-01-14 22:25:32 - r - INFO: - Episode: 36/100, Reward: 70.000, Step: 70 2023-01-14 22:25:32 - r - INFO: - Episode: 37/100, Reward: 65.000, Step: 65 2023-01-14 22:25:32 - r - INFO: - Episode: 38/100, Reward: 44.000, Step: 44 2023-01-14 22:25:33 - r - INFO: - Episode: 39/100, Reward: 99.000, Step: 99 2023-01-14 22:25:33 - r - INFO: - Episode: 40/100, Reward: 106.000, Step: 106 2023-01-14 22:25:34 - r - INFO: - Current episode 40 has the best eval reward: 130.100 2023-01-14 22:25:34 - r - INFO: - Episode: 41/100, Reward: 158.000, Step: 158 2023-01-14 22:25:35 - r - INFO: - Episode: 42/100, Reward: 44.000, Step: 44 2023-01-14 22:25:35 - r - INFO: - Episode: 43/100, Reward: 66.000, Step: 66 2023-01-14 22:25:35 - r - INFO: - Episode: 44/100, Reward: 43.000, Step: 43 2023-01-14 22:25:36 - r - INFO: - Episode: 45/100, Reward: 86.000, Step: 86 2023-01-14 22:25:36 - r - INFO: - Episode: 46/100, Reward: 71.000, Step: 71 2023-01-14 22:25:37 - r - INFO: - Episode: 47/100, Reward: 200.000, Step: 200 2023-01-14 22:25:38 - r - INFO: - Episode: 48/100, Reward: 200.000, Step: 200 2023-01-14 22:25:39 - r - INFO: - Episode: 49/100, Reward: 190.000, Step: 190 2023-01-14 22:25:39 - r - INFO: - Episode: 50/100, Reward: 200.000, Step: 200 2023-01-14 22:25:40 - r - INFO: - Current episode 50 has the best eval reward: 200.000 2023-01-14 22:25:41 - r - INFO: - Episode: 51/100, Reward: 200.000, Step: 200 2023-01-14 22:25:42 - r - INFO: - Episode: 52/100, Reward: 200.000, Step: 200 2023-01-14 22:25:43 - r - INFO: - Episode: 53/100, Reward: 200.000, Step: 200 2023-01-14 22:25:44 - r - INFO: - Episode: 54/100, Reward: 200.000, Step: 200 2023-01-14 22:25:45 - r - INFO: - Episode: 55/100, Reward: 200.000, Step: 200 2023-01-14 22:25:45 - r - INFO: - Current episode 55 has the best eval reward: 200.000 2023-01-14 22:25:46 - r - INFO: - Episode: 56/100, Reward: 200.000, Step: 200 2023-01-14 22:25:47 - r - INFO: - Episode: 57/100, Reward: 200.000, Step: 200 2023-01-14 22:25:48 - r - INFO: - Episode: 58/100, Reward: 200.000, Step: 200 2023-01-14 22:25:49 - r - INFO: - Episode: 59/100, Reward: 200.000, Step: 200 2023-01-14 22:25:50 - r - INFO: - Episode: 60/100, Reward: 200.000, Step: 200 2023-01-14 22:25:50 - r - INFO: - Current episode 60 has the best eval reward: 200.000 2023-01-14 22:25:51 - r - INFO: - Episode: 61/100, Reward: 200.000, Step: 200 2023-01-14 22:25:52 - r - INFO: - Episode: 62/100, Reward: 200.000, Step: 200 2023-01-14 22:25:53 - r - INFO: - Episode: 63/100, Reward: 200.000, Step: 200 2023-01-14 22:25:54 - r - INFO: - Episode: 64/100, Reward: 200.000, Step: 200 2023-01-14 22:25:55 - r - INFO: - Episode: 65/100, Reward: 200.000, Step: 200 2023-01-14 22:25:56 - r - INFO: - Current episode 65 has the best eval reward: 200.000 2023-01-14 22:25:56 - r - INFO: - Episode: 66/100, Reward: 200.000, Step: 200 2023-01-14 22:25:57 - r - INFO: - Episode: 67/100, Reward: 200.000, Step: 200 2023-01-14 22:25:58 - r - INFO: - Episode: 68/100, Reward: 200.000, Step: 200 2023-01-14 22:25:59 - r - INFO: - Episode: 69/100, Reward: 200.000, Step: 200 2023-01-14 22:26:00 - r - INFO: - Episode: 70/100, Reward: 200.000, Step: 200 2023-01-14 22:26:02 - r - INFO: - Episode: 71/100, Reward: 200.000, Step: 200 2023-01-14 22:26:02 - r - INFO: - Episode: 72/100, Reward: 200.000, Step: 200 2023-01-14 22:26:03 - r - INFO: - Episode: 73/100, Reward: 200.000, Step: 200 2023-01-14 22:26:04 - r - INFO: - Episode: 74/100, Reward: 200.000, Step: 200 2023-01-14 22:26:05 - r - INFO: - Episode: 75/100, Reward: 200.000, Step: 200 2023-01-14 22:26:07 - r - INFO: - Episode: 76/100, Reward: 200.000, Step: 200 2023-01-14 22:26:08 - r - INFO: - Episode: 77/100, Reward: 200.000, Step: 200 2023-01-14 22:26:08 - r - INFO: - Episode: 78/100, Reward: 200.000, Step: 200 2023-01-14 22:26:09 - r - INFO: - Episode: 79/100, Reward: 200.000, Step: 200 2023-01-14 22:26:10 - r - INFO: - Episode: 80/100, Reward: 200.000, Step: 200 2023-01-14 22:26:11 - r - INFO: - Current episode 80 has the best eval reward: 200.000 2023-01-14 22:26:12 - r - INFO: - Episode: 81/100, Reward: 200.000, Step: 200 2023-01-14 22:26:13 - r - INFO: - Episode: 82/100, Reward: 200.000, Step: 200 2023-01-14 22:26:14 - r - INFO: - Episode: 83/100, Reward: 200.000, Step: 200 2023-01-14 22:26:15 - r - INFO: - Episode: 84/100, Reward: 200.000, Step: 200 2023-01-14 22:26:15 - r - INFO: - Episode: 85/100, Reward: 200.000, Step: 200 2023-01-14 22:26:16 - r - INFO: - Current episode 85 has the best eval reward: 200.000 2023-01-14 22:26:17 - r - INFO: - Episode: 86/100, Reward: 200.000, Step: 200 2023-01-14 22:26:18 - r - INFO: - Episode: 87/100, Reward: 200.000, Step: 200 2023-01-14 22:26:19 - r - INFO: - Episode: 88/100, Reward: 200.000, Step: 200 2023-01-14 22:26:20 - r - INFO: - Episode: 89/100, Reward: 200.000, Step: 200 2023-01-14 22:26:21 - r - INFO: - Episode: 90/100, Reward: 200.000, Step: 200 2023-01-14 22:26:22 - r - INFO: - Current episode 90 has the best eval reward: 200.000 2023-01-14 22:26:22 - r - INFO: - Episode: 91/100, Reward: 198.000, Step: 198 2023-01-14 22:26:23 - r - INFO: - Episode: 92/100, Reward: 200.000, Step: 200 2023-01-14 22:26:24 - r - INFO: - Episode: 93/100, Reward: 200.000, Step: 200 2023-01-14 22:26:25 - r - INFO: - Episode: 94/100, Reward: 200.000, Step: 200 2023-01-14 22:26:26 - r - INFO: - Episode: 95/100, Reward: 200.000, Step: 200 2023-01-14 22:26:27 - r - INFO: - Current episode 95 has the best eval reward: 200.000 2023-01-14 22:26:28 - r - INFO: - Episode: 96/100, Reward: 200.000, Step: 200 2023-01-14 22:26:28 - r - INFO: - Episode: 97/100, Reward: 200.000, Step: 200 2023-01-14 22:26:29 - r - INFO: - Episode: 98/100, Reward: 200.000, Step: 200 2023-01-14 22:26:30 - r - INFO: - Episode: 99/100, Reward: 200.000, Step: 200 2023-01-14 22:26:31 - r - INFO: - Episode: 100/100, Reward: 200.000, Step: 200 2023-01-14 22:26:32 - r - INFO: - Current episode 100 has the best eval reward: 200.000 2023-01-14 22:26:32 - r - INFO: - Finish training!