-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathrun.py
149 lines (124 loc) · 5.73 KB
/
run.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
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
import argparse
import os
import logger
import yaml
import torch
try:
from yaml import CLoader as Loader, CDumper as Dumper
except ImportError:
from yaml import Loader, Dumper
from experiment_tools.factory import setup_params
# from algs.ppo_ltl import run_ppo_ltl
from algs.Q_learning import run_Q_learning
from algs.ppo import run_PPO
import numpy as np
from envs.abstract_env import Simulator
from automaton import Automaton, AutomatonRunner
import warnings
warnings.filterwarnings("ignore", category=UserWarning, module="torch.masked.maskedtensor")
def main(seed, param, to_redo):
automaton = AutomatonRunner(Automaton(**param['ltl']))
logger.info('*'*20 + '\tLTL: %s' % automaton.automaton.formula)
dir = os.path.join(param['logger']['dir_name'], 'ours_q', 'experiment_%05.f' % (seed) )
if ('discrete' in param['classes']) and (to_redo or not os.path.exists(os.path.join(os.getcwd(), 'experiments', dir))):
torch.manual_seed(seed)
np.random.seed(seed)
param['env']['file'] = param['classes']['discrete']
env = setup_params(param)
sim = Simulator(env, automaton)
# Simple check to see if observation_space space is discrete
is_discrete_obs_space = False
try:
sim.observation_space['mdp'].n
is_discrete_obs_space = True
except:
pass
logger.configure(name=dir)
run_Q_learning(param, sim, False, not is_discrete_obs_space, to_hallucinate=True)
dir = os.path.join(param['logger']['dir_name'], 'baseline_q', 'experiment_%05.f' % (seed) )
if ('discrete' in param['classes']) and (to_redo or not os.path.exists(os.path.join(os.getcwd(), 'experiments', dir))):
torch.manual_seed(seed)
np.random.seed(seed)
param['env']['file'] = param['classes']['discrete']
env = setup_params(param)
sim = Simulator(env, automaton)
# Simple check to see if observation_space space is discrete
is_discrete_obs_space = False
try:
sim.observation_space['mdp'].n
is_discrete_obs_space = True
except:
pass
logger.configure(name=dir)
run_Q_learning(param, sim, False, not is_discrete_obs_space, to_hallucinate=False)
dir = os.path.join(param['logger']['dir_name'], 'ours_ppo', 'experiment_%05.f' % (seed) )
if ('continuous' in param['classes']) and (to_redo or not os.path.exists(os.path.join(os.getcwd(), 'experiments', dir))):
torch.manual_seed(seed)
np.random.seed(seed)
param['env']['file'] = param['classes']['continuous']
env = setup_params(param)
sim = Simulator(env, automaton)
# Simple check to see if observation_space space is discrete
is_discrete_obs_space = False
try:
sim.observation_space['mdp'].n
is_discrete_obs_space = True
# PPO for discrete state space: NOT IMPLEMENTED
return
except:
pass
logger.configure(name=dir)
run_PPO(param, sim, False, not is_discrete_obs_space, to_hallucinate=True)
dir = os.path.join(param['logger']['dir_name'], 'baseline_ppo', 'experiment_%05.f' % (seed) )
if ('continuous' in param['classes']) and (to_redo or not os.path.exists(os.path.join(os.getcwd(), 'experiments', dir))):
torch.manual_seed(seed)
np.random.seed(seed)
param['env']['file'] = param['classes']['continuous']
env = setup_params(param)
sim = Simulator(env, automaton)
# Simple check to see if observation_space space is discrete
is_discrete_obs_space = False
try:
sim.observation_space['mdp'].n
is_discrete_obs_space = True
# PPO for discrete state space: NOT IMPLEMENTED
return
except:
pass
logger.configure(name=dir)
run_PPO(param, sim, False, not is_discrete_obs_space, to_hallucinate=False)
# dir = os.path.join(param['logger']['dir_name'], 'baseline_only_init', 'experiment_%05.f' % (seed) )
# if to_redo or not os.path.exists(os.path.join(os.getcwd(), 'experiments', dir)):
# logger.configure(name=dir)
# torch.manual_seed(seed)
# np.random.seed(seed)
# param['env']['file'] = param['classes']['continuous']
# env = setup_params(param)
# sim = Simulator(env, automaton)
# # Simple check to see if observation_space space is discrete
# is_discrete_obs_space = False
# try:
# sim.observation_space['mdp'].n
# is_discrete_obs_space = True
# except:
# pass
# run_ppo_continuous(param, sim, False, to_hallucinate=False)
if __name__ == '__main__':
# Local:
# python run.py chain.yaml
parser = argparse.ArgumentParser(description='Run Experiment')
parser.add_argument('cfg', help='config file', type=str)
parser.add_argument('-r', '--restart', action='store_true')
args = parser.parse_args()
assert args.cfg.endswith('.yaml'), 'Must be yaml file'
with open(os.getcwd() + '/cfgs/{0}'.format(args.cfg), 'r') as f:
param = yaml.load(f, Loader=Loader)
np.random.seed(param['init_seed'])
seeds = [np.random.randint(1e6) for _ in range(param['n_seeds'])]
for seed in seeds:
print('*' * 20)
# param['logger']['name'] = #'experts_and_'+param['MCTS']['bandit_strategy'] if param['experiment']['experts'] else 'policy_and_'+param['MCTS']['bandit_strategy']
# logger.configure(name=os.path.join(param['logger']['dir_name'], 'experiment_%05.f' % (seed) ))
logger.Logger.set_level(logger,logger.DEBUG)
logger.info("Seed = {}".format(float(seed)))
main(seed, param, args.restart)