-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_dense.yaml
174 lines (170 loc) · 3.13 KB
/
test_dense.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
167
168
169
170
171
172
173
174
# Please update env.dataset to dataset path
env:
venv_seed: 42
name: advanced-dense
dataset: data/advanced_dense_test_700.pkl
action_dim: 2
action_mag: 1.0
use_append: true
obs_buffer: 0.25
g_x_fail: 0.25
terminal_type: const
img_h: 90
img_w: 160
max_step_train: 200
max_step_eval: 200
obs_channel: 12
fixed_init: true
sparse_reward: false
num_env_train: 1000
reward_type: task
reward_goal: 5
reward_wander: 1
reward_obs: -10
#
mesh_parent_path: data/3D-FUTURE-model
task_parent_path: data/3d-front-tasks-advanced-dense
fill_obs: false
camera_tilt: 15
camera_tilt_range: [15, 25]
camera_roll_noise_std: 3
perf_xdot_range: [0.5, 1.0]
backup_xdot_range: [0.2, 0.5]
device: cuda:0
traj_size: 4
frame_skip: 3
shield_dict:
type: value
threshold: -0.05
performance:
train:
eval: true
device: cuda:0
batch_size: 128
update_period: 1
alpha: 0.5
learn_alpha: false
gamma: 0.99
gamma_schedule: false
lr_a: 0.001
lr_al: 0.001
lr_c: 0.001
lr_a_schedule: false
lr_c_schedule: false
mode: performance
tau: 0.01
terminal_type: none
actor_path: checkpoint/advanced/pac_perf/perfactor.pth
critic_path: checkpoint/advanced/pac_perf/perf-critic.pth
# latent
lr_d: 0.001
disc_batch_size: 32
disc_recent_size: 1000
aug_reward_range: 1
aug_reward_range_schedule: False
aug_reward_range_init: 0.0
aug_reward_range_end: 1.0
aug_reward_range_step: 1.0
aug_reward_range_period: 1500
latent_dim: 20
latent_dim_cnn: 10
latent_std_schedule: False
latent_prior_std: 0.5
latent_std_init: 0.5
latent_std_end: 2.0
latent_std_step: 0.5
latent_std_min_step: 0
latent_std_period: 5000
arch:
critic_has_act_ind: false
act_ind:
- 1
activation:
actor: ReLU
critic: ReLU
kernel_size:
- 7
- 5
- 3
stride:
- 4
- 3
- 2
n_channel:
- 16
- 32
- 64
mlp_dim:
actor:
- 256
- 256
- 256
critic:
- 256
- 256
- 256
disc:
- 256
- 256
append_dim: 2
use_bn: false
use_ln: true
use_sm: false
use_spec_disc: true
backup:
train:
eval: true
device: cuda:0
alpha: 0.01
learn_alpha: true
batch_size: 128
gamma: 0.8
gamma_decay: 0.5
gamma_end: 0.995
gamma_period: 400000
gamma_schedule: true
lr_a: 0.001
lr_al: 0.0001
lr_c: 0.001
lr_a_schedule: false
lr_c_schedule: false
mode: safety
tau: 0.01
terminal_type: none
actor_path: checkpoint/advanced/pac_perf/backup-actor.pth
critic_path: checkpoint/advanced/pac_perf/backup-critic.pth
# latent
latent_dim: 0
latent_dim_cnn: 0
arch:
critic_has_act_ind: true
act_ind:
- -1
activation:
actor: ReLU
critic: ReLU
kernel_size:
- 7
- 5
- 3
stride:
- 4
- 3
- 2
n_channel:
- 16
- 32
- 64
mlp_dim:
actor:
- 256
- 256
- 256
critic:
- 256
- 256
- 256
append_dim: 2
use_bn: false
use_ln: true
use_sm: false