Update README.md
Browse files
README.md
CHANGED
@@ -12,12 +12,49 @@ configs:
|
|
12 |
data_files:
|
13 |
- "atari_battle_zone_ppo.csv"
|
14 |
- "atari_double_dunk_ppo.csv"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
- config_name: dqn_data
|
16 |
data_files:
|
17 |
- "atari_battle_zone_dqn.csv"
|
18 |
- "atari_double_dunk_dqn.csv"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
- config_name: sac_data
|
20 |
-
data_files:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
---
|
22 |
|
23 |
# The ARLBench Performance Dataset
|
@@ -26,5 +63,5 @@ configs:
|
|
26 |
Since we performed several thousand runs on the benchmark to find meaningful HPO test settings in RL, we collect them in this dataset for future use.
|
27 |
These runs could be used to meta-learn information about the hyperparameter landscape or warmstart HPO tools.
|
28 |
|
29 |
-
In detail, it contains each 10 runs for PPO, DQN and SAC respectively on the Atari-5 environments, four XLand gridworlds, four
|
30 |
For more information, refer to the ARLBench paper.
|
|
|
12 |
data_files:
|
13 |
- "atari_battle_zone_ppo.csv"
|
14 |
- "atari_double_dunk_ppo.csv"
|
15 |
+
- "atari_phoenix_ppo.csv"
|
16 |
+
- "atari_qbert_ppo.csv"
|
17 |
+
- "atari_this_game_ppo.csv"
|
18 |
+
- "box2d_lunar_lander_continuous_ppo.csv"
|
19 |
+
- "box2d_lunar_lander_ppo.csv"
|
20 |
+
- "brax_halfcheetah_ppo.csv"
|
21 |
+
- "brax_hopper_ppo.csv"
|
22 |
+
- "brax_ant_ppo.csv"
|
23 |
+
- "brax_humanoid_ppo.csv"
|
24 |
+
- "cc_acrobot_ppo.csv"
|
25 |
+
- "cc_cartpole_ppo.csv"
|
26 |
+
- "cc_continuous_mountain_car_ppo.csv"
|
27 |
+
- "cc_mountain_car_ppo.csv"
|
28 |
+
- "cc_pendulum_ppo.csv"
|
29 |
+
- "minigrid_door_key_ppo.csv"
|
30 |
+
- "minigrid_empty_random_ppo.csv"
|
31 |
+
- "minigrid_four_rooms_ppo.csv"
|
32 |
+
- "minigrid_unlock_ppo.csv"
|
33 |
- config_name: dqn_data
|
34 |
data_files:
|
35 |
- "atari_battle_zone_dqn.csv"
|
36 |
- "atari_double_dunk_dqn.csv"
|
37 |
+
- "atari_phoenix_dqn.csv"
|
38 |
+
- "atari_qbert_dqn.csv"
|
39 |
+
- "atari_this_game_dqn.csv"
|
40 |
+
- "minigrid_door_key_dqn.csv"
|
41 |
+
- "minigrid_empty_random_dqn.csv"
|
42 |
+
- "minigrid_four_rooms_dqn.csv"
|
43 |
+
- "minigrid_unlock_dqn.csv"
|
44 |
+
- "cc_acrobot_dqn.csv"
|
45 |
+
- "cc_cartpole_dqn.csv"
|
46 |
+
- "cc_mountain_car_dqn.csv"
|
47 |
+
- "box2d_lunar_lander_dqn.csv"
|
48 |
- config_name: sac_data
|
49 |
+
data_files:
|
50 |
+
- "box2d_bipedal_walker_sac.csv"
|
51 |
+
- "box2d_lunar_lander_continuous_sac.csv"
|
52 |
+
- "brax_halfcheetah_sac.csv"
|
53 |
+
- "brax_hopper_sac.csv"
|
54 |
+
- "brax_ant_sac.csv"
|
55 |
+
- "brax_humanoid_sac.csv"
|
56 |
+
- "cc_continuous_mountain_car_sac.csv"
|
57 |
+
- "cc_pendulum_sac.csv"
|
58 |
---
|
59 |
|
60 |
# The ARLBench Performance Dataset
|
|
|
63 |
Since we performed several thousand runs on the benchmark to find meaningful HPO test settings in RL, we collect them in this dataset for future use.
|
64 |
These runs could be used to meta-learn information about the hyperparameter landscape or warmstart HPO tools.
|
65 |
|
66 |
+
In detail, it contains each 10 runs for PPO, DQN and SAC respectively on the Atari-5 environments, four XLand gridworlds, four Brax walkers, five classic control and two Box2D environments.
|
67 |
For more information, refer to the ARLBench paper.
|