PaintRL/param_test_apex_dqn.py at master · translearn/PaintRL · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
from param_test_ppo import main


if __name__ == '__main__':
    configuration = {
        'num_workers': 15,
        # 'num_envs_per_worker': 4,
        'num_gpus': 1,

        'num_atoms': 1,
        'dueling': False,
        'double_q': True,
        'hiddens': [64],

        'n_step': 3,
        'lr': 0.0001,
        'adam_epsilon': .00015,

        'timesteps_per_iteration': 2000,
        'target_network_update_freq': 5000,
        'soft_q': False,
        'parameter_noise': False,
        'batch_mode': 'truncate_episodes',

        'schedule_max_timesteps': 200000,
        'exploration_final_eps': 0.01,
        'exploration_fraction': .1,
        'prioritized_replay_alpha': 0.5,
        'beta_annealing_fraction': 1.0,
        'final_prioritized_replay_beta': 1.0,
        'buffer_size': 100000,
        'prioritized_replay': True,

        # 'learning_starts': 1000,
        'sample_batch_size': 20,
        'train_batch_size': 512,

        'gamma': 1,
    }
    main('APEX', configuration)