-
Notifications
You must be signed in to change notification settings - Fork 209
/
Copy pathtrpo.yml
174 lines (161 loc) · 3.18 KB
/
trpo.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
# Note: OpenAI uses 16 mpi workers for training on atari
# in practice it is run with with 4 workers
atari:
n_timesteps: !!float 1e7
policy: 'CnnPolicy'
timesteps_per_batch: 512
max_kl: 0.001
cg_iters: 10
cg_damping: !!float 1e-3
entcoeff: 0.01
gamma: 0.98
lam: 1
vf_iters: 3
vf_stepsize: !!float 1e-4
# Tuned
Pendulum-v0:
n_timesteps: !!float 3e5
policy: 'MlpPolicy'
max_kl: 0.000193
timesteps_per_batch: 1024
gamma: 0.99
lam: 0.9
entcoeff: 0.01118
cg_damping: 2.35e-05
vf_iters: 10
vf_stepsize: 0.00428
MountainCar-v0:
n_timesteps: !!float 1e6
policy: 'MlpPolicy'
# Tuned
MountainCarContinuous-v0:
n_timesteps: !!float 1.2e5
policy: 'MlpPolicy'
max_kl: 1.77e-06
timesteps_per_batch: 32
entcoeff: 0.0
cg_iters: 20
lam: 0.95
gamma: 0.9999
cg_damping: 0.0002
vf_iters: 10
vf_stepsize: !!float 0.0463
CartPole-v1:
n_timesteps: !!float 1e6
policy: 'MlpPolicy'
timesteps_per_batch: 512
max_kl: 0.001
cg_iters: 10
cg_damping: !!float 1e-3
entcoeff: 0.0
gamma: 0.99
lam: 1
vf_iters: 3
vf_stepsize: !!float 1e-4
LunarLander-v2:
n_timesteps: !!float 2e6
policy: 'MlpPolicy'
timesteps_per_batch: 1024
max_kl: 0.01
cg_iters: 10
cg_damping: 0.1
entcoeff: 0.0
gamma: 0.99
lam: 0.98
vf_iters: 5
vf_stepsize: !!float 1e-3
LunarLanderContinuous-v2:
n_timesteps: !!float 1e6
policy: 'MlpPolicy'
timesteps_per_batch: 1024
max_kl: 0.01
cg_iters: 10
cg_damping: 0.1
entcoeff: 0.0
gamma: 0.99
lam: 0.98
vf_iters: 5
vf_stepsize: !!float 1e-3
# Tuned
HalfCheetahBulletEnv-v0:
env_wrapper: utils.wrappers.TimeFeatureWrapper
n_timesteps: !!float 2e6
policy: 'MlpPolicy'
timesteps_per_batch: 2048
max_kl: 0.01
cg_iters: 15
cg_damping: 0.1
entcoeff: 0.0
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3
# Tuned
AntBulletEnv-v0:
env_wrapper: utils.wrappers.TimeFeatureWrapper
n_timesteps: !!float 2e6
policy: 'MlpPolicy'
timesteps_per_batch: 2048
max_kl: 0.01
cg_iters: 15
cg_damping: 0.1
entcoeff: 0.0
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3
# Tuned
Walker2DBulletEnv-v0:
env_wrapper: utils.wrappers.TimeFeatureWrapper
n_timesteps: !!float 2e6
policy: 'MlpPolicy'
timesteps_per_batch: 2048
max_kl: 0.01
cg_iters: 20
cg_damping: 0.1
entcoeff: 0.001
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3
# Tuned
HopperBulletEnv-v0:
env_wrapper: utils.wrappers.TimeFeatureWrapper
n_timesteps: !!float 2e6
policy: 'MlpPolicy'
timesteps_per_batch: 2048
max_kl: 0.005
cg_iters: 15
cg_damping: 0.1
entcoeff: 0.01
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3
# Tuned
BipedalWalker-v3:
env_wrapper: utils.wrappers.TimeFeatureWrapper
n_timesteps: !!float 5e6
policy: 'MlpPolicy'
timesteps_per_batch: 2048
max_kl: 0.01
cg_iters: 15
cg_damping: 0.1
entcoeff: 0.001
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3
# To be tuned
BipedalWalkerHardcore-v3:
n_timesteps: !!float 5e7
policy: 'MlpPolicy'
timesteps_per_batch: 4096
max_kl: 0.02
cg_iters: 20
cg_damping: 0.1
entcoeff: 0.001
gamma: 0.99
lam: 0.95
vf_iters: 5
vf_stepsize: !!float 1e-3