forked from qgallouedec/panda-gym
-
Notifications
You must be signed in to change notification settings - Fork 0
/
wandb_sweep_config.yaml
166 lines (165 loc) · 3.03 KB
/
wandb_sweep_config.yaml
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
program: wandb_train.py
method: bayes
metric:
name: global_step
goal: minimize
parameters:
tau:
distribution: uniform
min: 0.005
max: 0.04
gamma:
distribution: uniform
min: 0.49
max: 0.99
use_sde:
distribution: categorical
values:
- True
- False
ent_coef:
distribution: categorical
values:
- auto
env_name:
distribution: categorical
values:
- PandaReachAO-v3
algorithm:
distribution: categorical
values:
- TQC
batch_size:
distribution: int_uniform
min: 64
max: 512
n_substeps:
distribution: int_uniform
min: 2
max: 50
train_freq:
distribution: int_uniform
min: 4
max: 8
buffer_size:
distribution: int_uniform
min: 150000
max: 1000000
policy_type:
distribution: categorical
values:
- MultiInputPolicy
reward_type:
distribution: categorical
values:
- sparse
action_noise:
distribution: categorical
values:
- None
control_type:
distribution: categorical
values:
- js
policy_class:
distribution: categorical
values:
- <class 'sb3_contrib.tqc.policies.MultiInputPolicy'>
learning_rate:
distribution: uniform
min: 0.00035
max: 0.0015
max_timesteps:
distribution: categorical
values:
- 300_000
max_ep_steps:
distribution: categorical
values:
- [50]
- [100]
- [200]
- [400]
policy_kwargs.log_std_init:
distribution: int_uniform
min: -6
max: -1
action_limiter:
distribution: categorical
values:
- clip
gradient_steps:
distribution: int_uniform
min: 4
max: 8
target_entropy:
distribution: categorical
values:
- "auto"
learning_starts:
distribution: categorical
values:
- 10000
show_goal_space:
distribution: categorical
values:
- False
n_envs:
distribution: categorical
values:
- 1
goal_distance_threshold:
distribution: categorical
values:
- 0.05
show_debug_labels:
distribution: categorical
values:
- False
joint_obstacle_observation:
distribution: categorical
values:
- vectors
top_quantiles_to_drop_per_net:
distribution: int_uniform
min: 1
max: 4
stages:
distribution: categorical
values:
- ["wang_3"]
prior_steps:
distribution: categorical
values:
- 0
obs_type:
distribution: categorical
values:
- [ "ee","js" ]
randomize_robot_pose:
distribution: categorical
values:
- False
- True
truncate_episode_on_collision:
distribution: categorical
values:
- True
policy_kwargs:
distribution: categorical
values:
- {net_arch: [400, 300]}
- {net_arch: [256, 256]}
- {net_arch: [256, 128]}
- {net_arch: [512, 256]}
- {net_arch: [512, 512]}
- {net_arch: [256, 256, 256]}
collision_reward:
distribution: int_uniform
min: -25
max: -500
command:
- ${env}
- venv/Scripts/python
- ${program}
- ${args}