-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathlauncher_gsac2.py
135 lines (112 loc) · 4.4 KB
/
launcher_gsac2.py
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
import os
from traitlets import default
try:
os.remove(os.path.expanduser("~/miniconda/envs/py37/lib/python3.7/site-packages/mujoco_py/generated/mujocopy-buildlock.lock"))
except:
pass
from spinup.utils.run_utils import ExperimentGrid
from spinup import gsac2_pytorch
import torch
from multiprocessing import Pool
def expr(args, seed):
test_trajs_names = {}
test_trajs_names['Walker2d-v3'] = './data/Walker2d-v3_sac_base_50_trajs.pkl'
test_trajs_names['Hopper-v3'] = './data/Hopper-v3_sac_base_50_trajs.pkl'
test_trajs_names['HalfCheetah-v3'] = './data/HalfCheetah-v3_sac_base_50_trajs.pkl'
test_trajs_names['Ant-v3'] = './data/Ant-v3_sac_base_50_trajs.pkl'
test_trajs_names['Humanoid-v3'] = './data/Humanoid-v3_sac_base_50_trajs.pkl'
name = args.env + "_" + args.name
eg = ExperimentGrid(name)
eg.add('env_name', args.env)
eg.add('seed', [seed])
eg.add('epochs', args.epochs)
eg.add('steps_per_epoch', 4000)
eg.add('ac_kwargs:hidden_sizes', [(256, 256)], 'hid')
eg.add('ac_kwargs:activation', [torch.nn.ReLU], '')
eg.add('test_trajs_name', test_trajs_names[args.env])
eg.add('train_trajs_top_ratio', args.train_trajs_top_ratio)
eg.add('trajs_sample_ratio', args.trajs_sample_ratio)
eg.add('sample_num', args.sample_num)
eg.add('continue_step', args.continue_step)
eg.add('sample_rule', args.sample_rule)
eg.add('over_write_ratio', args.over_write_ratio)
eg.run(gsac2_pytorch, num_cpu=args.cpu)
def all_expr(args, seed):
if args.env == 'three' or args.env == 'all':
args.env = 'Walker2d-v3'
expr(args,seed)
args.env = 'Humanoid-v3'
expr(args,seed)
args.env = 'Hopper-v3'
expr(args,seed)
if args.env == 'all':
args.env = 'Ant-v3'
expr(args,seed)
args.env = 'HalfCheetah-v3'
expr(args,seed)
else:
expr(args,seed)
def set_default_and_name(args):
name = args.name
if args.epochs == None:
if args.env == 'Hopper-v3':
args.epochs = 265
else:
args.epochs = 800
if args.train_trajs_top_ratio == None:
args.train_trajs_top_ratio = 0.5
else:
name += '_tttr' + (str(int(args.train_trajs_top_ratio*100)))
if args.trajs_sample_ratio == None:
args.trajs_sample_ratio = 0.5
else:
name += '_tsr' + (str(int(args.trajs_sample_ratio * 10)))
if args.sample_num == None:
args.sample_num = 1
else:
name += '_sn' + (str(args.sample_num))
if args.continue_step == None:
args.continue_step = 100
else:
name += '_cs' + (str(args.continue_step))
if args.sample_rule == None:
args.sample_rule = 0
else:
name += '_sr' + (str(args.sample_rule))
if args.over_write_ratio == None:
args.over_write_ratio = 1.
else:
name += '_owr' + (str( int(args.over_write_ratio*10) ))
if args.start_sample_ratio == None:
args.start_sample_ratio = 0.5
else:
name += "_ssr" + (str(int(args.start_sample_ratio*100)))
args.name = name
return name
if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--cpu', type=int, default=1)
parser.add_argument('--num_runs', type=int, default=5)
parser.add_argument('--env', type=str, default='Walker2d-v3')
parser.add_argument('--name', type=str, default='gsac3')
parser.add_argument('--seed', type=int, default=1324)
parser.add_argument('--epochs', type=int, default=900)
parser.add_argument('--train_trajs_top_ratio', '-tttr', type=float)
parser.add_argument('--trajs_sample_ratio', '-tsr', type=float)
parser.add_argument('--sample_num', '-sn', type=int)
parser.add_argument('--continue_step', '-cs', type=int)
parser.add_argument('--sample_rule', '-sr', type=int)
parser.add_argument('--over_write_ratio', '-owr', type=float)
parser.add_argument('--start_sample_ratio', '-ssr', type=float)
args = parser.parse_args()
set_default_and_name(args)
print(args)
print("name ", args.name)
if args.num_runs > 1:
# start multiprocessing only if more than one runs
x = args.seed
with Pool(args.num_runs) as p:
p.starmap(all_expr, [(args, seed) for seed in range(x, x + args.num_runs)])
else:
all_expr(args, args.seed)