-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathexperiments.yml
345 lines (315 loc) · 6.92 KB
/
experiments.yml
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
Baseline:
LUNARLANDER:
gym_id: mo-LunarLanderContinuous-v2
num_envs: 16
total_timesteps: 100000
HOPPER:
gym_id: mo-Hopper-v4
num_envs: 16
total_timesteps: 500000
PENDULUM:
gym_id: mo-Pendulum-v1
num_envs: 16
total_timesteps: 50000
HUMANOID:
gym_id: mo-Humanoid-v4
num_envs: 16
total_timesteps: 500000
HALFCHEETAH:
gym_id: mo-HalfCheetah-v4
num_envs: 16
total_timesteps: 500000
HALFCHEETAHEF:
gym_id: mo-HalfCheetahEF-v4
num_envs: 16
total_timesteps: 500000
VSS:
gym_id: mo-VSS-v0
num_envs: 20
total_timesteps: 500000
learning_starts: 5000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
VSSEF:
gym_id: mo-VSSEF-v0
num_envs: 20
total_timesteps: 500000
learning_starts: 5000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
TAXI:
gym_id: mo-Taxi-v3
total_episodes: 2000
gamma: 0.9999
q_lr: 0.1
epsilon: 0.8
QLUNARLANDER:
gym_id: mo-LunarLander-v2
num_envs: 16
total_timesteps: 100000
epsilon: 0.8
MINECART:
gym_id: mo-Minecart-v0
num_envs: 16
total_timesteps: 200000
gamma: 0.98
target_network_frequency: 200
tau: 0.00001
epsilon: 1.0
epsilon_decay_factor: 0.99997
Dylam:
LUNARLANDER:
gym_id: mo-LunarLanderContinuous-v2
num_envs: 16
total_timesteps: 100000
stratified: True
dylam: True
num_rewards: 4
r_max: [1000, -100, -20, 1]
r_min: [-1000, -1000, -1000, -1]
dylam_rb: 100
comp_names: ["Shaping", "Power_linear", "Power_angular", "Goal"]
HOPPER:
gym_id: mo-Hopper-v4
num_envs: 16
total_timesteps: 500000
stratified: True
dylam: True
num_rewards: 4
r_max: [1000, 1000, -100, 1000]
r_min: [0, -200, -250, 0]
dylam_rb: 100
dylam_tau: 0.9999
comp_names: ["Forward", "Jump", "Energy", "Healthy"]
PENDULUM:
gym_id: mo-Pendulum-v1
num_envs: 16
total_timesteps: 50000
stratified: True
dylam: True
num_rewards: 3
r_max: [0, 0, 0]
r_min: [-100, -100, -100]
dylam_rb: 100
dylam_tau: 0.9999
comp_names: ["Theta", "Angular_vel", "Torque"]
HUMANOID:
gym_id: mo-Humanoid-v4
num_envs: 16
total_timesteps: 500000
stratified: True
dylam: True
num_rewards: 3
r_max: [1000, 0, 1000]
r_min: [0, -1000, 0]
dylam_rb: 100
dylam_tau: 0.9999
comp_names: ["Forward", "Energy", "Healthy"]
HALFCHEETAH:
gym_id: mo-HalfCheetah-v4
num_envs: 16
total_timesteps: 500000
stratified: True
dylam: True
num_rewards: 2
r_max: [800, -200]
r_min: [0, -800]
dylam_rb: 500
dylam_tau: 0.9999
comp_names: ["Run", "Control"]
HALFCHEETAHEF:
gym_id: mo-HalfCheetahEF-v4
num_envs: 16
total_timesteps: 500000
stratified: True
dylam: True
num_rewards: 2
r_max: [800, 800]
r_min: [0, 0]
dylam_rb: 500
dylam_tau: 0.9999
comp_names: ["Run", "Efficiency"]
VSS:
gym_id: mo-VSS-v0
num_envs: 20
total_timesteps: 500000
learning_starts: 5000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
stratified: True
dylam: True
num_rewards: 3
r_max: [150, 40, -100]
r_min: [0, 0, -300]
dylam_rb: 500
dylam_tau: 0.9999
comp_names: ["Move", "Ball", "Energy"]
VSSEF:
gym_id: mo-VSSEF-v0
num_envs: 20
total_timesteps: 500000
learning_starts: 5000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
stratified: True
dylam: True
num_rewards: 3
r_max: [70, 40, 30]
r_min: [0, 0, 0]
dylam_rb: 500
dylam_tau: 0.9999
comp_names: ["Move", "Ball", "Efficiency"]
TAXI:
gym_id: mo-Taxi-v3
total_episodes: 2000
gamma: 0.9999
q_lr: 0.1
epsilon: 0.8
stratified: True
num_rewards: 3
dylam: True
lambdas: [0.3333, 0.3333, 0.3333]
r_max: [-20, 1, 0]
r_min: [-200, 0, -10]
dylam_tau: 0.995
dylam_rb: 10
comp_names: ["Energy", "Objective", "Illegal_action"]
QLUNARLANDER:
gym_id: mo-LunarLander-v2
num_envs: 16
total_timesteps: 100000
epsilon: 0.8
stratified: True
dylam: True
num_rewards: 4
r_max: [1000, -100, -20, 1]
r_min: [-1000, -1000, -1000, -1]
dylam_rb: 100
dylam_tau: 0.9999
comp_names: ["Shaping", "Power_linear", "Power_angular", "Goal"]
MINECART:
gym_id: mo-Minecart-v0
num_envs: 16
total_timesteps: 200000
gamma: 0.98
target_network_frequency: 200
tau: 0.00001
epsilon: 1.0
epsilon_decay_factor: 0.99997
stratified: True
dylam: True
num_rewards: 3
r_max: [1.5, 1.5, 0]
r_min: [0, 0, -20]
comp_names: ["First_minerium", "Second_minerium", "Fuel"]
GPILS:
LUNARLANDER:
gym_id: mo-LunarLanderContinuous-v2
num_envs: 16
total_timesteps: 100000
stratified: True
num_rewards: 4
steps_per_iteration: 10000
num_eval_episodes: 3
HOPPER:
gym_id: mo-Hopper-v4
num_envs: 16
total_timesteps: 500000
stratified: True
num_rewards: 4
steps_per_iteration: 10000
reward_scaling: 1000
VSSEF:
gym_id: mo-VSSEF-v0
num_envs: 20
total_timesteps: 500000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
stratified: True
num_rewards: 3
steps_per_iteration: 10000
HALFCHEETAH:
gym_id: mo-HalfCheetah-v4
num_envs: 16
total_timesteps: 500000
num_rewards: 2
steps_per_iteration: 10000
HALFCHEETAHEF:
gym_id: mo-HalfCheetahEF-v4
num_envs: 16
total_timesteps: 500000
stratified: True
dylam: True
num_rewards: 2
steps_per_iteration: 10000
Drq:
TAXI:
gym_id: mo-Taxi-v3
total_episodes: 2000
epsilon: 0.8
gamma: 0.9999
q_lr: 0.1
stratified: True
num_rewards: 3
comp_names: ["Energy", "Objective", "Illegal_action"]
QLUNARLANDER:
gym_id: mo-LunarLander-v2
num_envs: 16
total_timesteps: 100000
epsilon: 0.8
stratified: True
num_rewards: 4
comp_names: ["Shaping", "Power_linear", "Power_angular", "Goal"]
HALFCHEETAH:
gym_id: mo-HalfCheetah-v4
num_envs: 16
total_timesteps: 500000
stratified: True
num_rewards: 2
comp_names: ["Run", "Control"]
VSS:
gym_id: mo-VSS-v0
num_envs: 20
total_timesteps: 500000
learning_starts: 5000
tau: 0.999
policy_frequency: 1
policy_lr: 0.0001
q_lr: 0.0001
batch_size: 256
n_hidden: 4
gamma: 0.95
update_frequency: 1
stratified: True
num_rewards: 3
comp_names: ["Move", "Ball", "Energy"]