-
Notifications
You must be signed in to change notification settings - Fork 11
/
Copy pathrlberry_params.yml
121 lines (117 loc) · 2.41 KB
/
rlberry_params.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
REINFORCE:
init_kwargs:
batch_size: 8
horizon: 256
gamma: 0.99
entr_coef: 0.01
learning_rate: 0.0001
normalize: True
optimizer_type: "ADAM"
policy_net_fn: null
policy_net_kwargs: null
use_bonus_if_available: False
device: "cuda:best"
fit_kwargs:
fit_budget: 10
eval_kwargs:
eval_horizon: 200
n_simulations: 3
gamma: 0.99
PPO:
init_kwargs:
batch_size: 64
n_steps: 2048
gamma: 0.99
entr_coef: 0.01
vf_coef: 0.5
learning_rate: 0.01
optimizer_type: "ADAM"
eps_clip: 0.2
k_epochs: 5
use_gae: True
gae_lambda: 0.95
policy_net_fn: null
value_net_fn: null
policy_net_kwargs: null
value_net_kwargs: null
normalize_rewards: False
normalize_advantages: False
device: "cuda:best"
fit_kwargs:
fit_budget: 1000
eval_kwargs:
eval_horizon: 200
n_simulations: 3
gamma: 0.99
A2C:
init_kwargs:
batch_size: 256
gamma: 0.99
entr_coef: 0.01
learning_rate: 0.01
optimizer_type: "ADAM"
policy_net_fn: null
value_net_fn: null
policy_net_kwargs: null
value_net_kwargs: null
device: "cuda:best"
eval_interval: null
fit_kwargs:
fit_budget: 1000
eval_kwargs:
eval_horizon: 200
n_simulations: 3
gamma: 0.99
DQN:
init_kwargs:
gamma: 0.99
batch_size: 32
chunk_size: 8
lambda_: 0.5
target_update_parameter: 0.005
device: "cuda:best"
learning_rate: 1.0e-3
epsilon_init: 1.0
epsilon_final: 0.1
epsilon_decay_interval: 20000
loss_function: "l2"
optimizer_type: "ADAM"
q_net_constructor: null
q_net_kwargs: null
use_double_dqn: False
use_prioritized_replay: False
train_interval: 10
gradient_steps: 1
max_replay_size: 200000
learning_starts: 5000
eval_interval: null
fit_kwargs:
fit_budget: 1000
eval_kwargs:
eval_horizon: 200
n_simulations: 3
gamma: 0.99
SAC:
init_kwargs:
batch_size: 8
gamma: 0.99
entr_coef: 0.01
learning_rate: 0.01
buffer_capacity: 30000
optimizer_type: "ADAM"
k_epochs: 5
policy_net_fn: null
value_net_fn: null
twinq_net_fn: null
policy_net_kwargs: null
value_net_kwargs: null
twinq_net_kwargs: null
use_bonus: False
uncertainty_estimator_kwargs: null
device: "cuda:best"
fit_kwargs:
fit_budget: 100
eval_kwargs:
eval_horizon: 200
n_simulations: 3
gamma: 0.99