def random_episodes(env_ctor, num_episodes, output_dir=None):
env = env_ctor()
env = wrappers.CollectGymDataset(env, output_dir)
episodes = []
for _ in range(num_episodes):
policy = lambda env, obs: env.action_space.sample()
done = False
obs = env.reset()
while not done:
action = policy(env, obs)
obs, _, done, info = env.step(action)
episodes.append(info['episode'])
return episodes
def random_episodes(env_ctor, num_episodes, outdir=None):
env = env_ctor()
env = wrappers.CollectGymDataset(env, outdir)
episodes = [] if outdir else None
for _ in range(num_episodes):
policy = lambda env, obs: env.action_space.sample()
done = False
obs = env.reset()
while not done:
action = policy(env, obs)
obs, _, done, info = env.step(action)
if outdir is None:
episodes.append(info['episode'])
try:
env.close()
except AttributeError:
pass
return episodes
def random_episodes(env_ctor, num_episodes, output_dir=None):
env = env_ctor() # env is an <ExternalProcess object>.
env = wrappers.CollectGymDataset(env, output_dir)
episodes = []
for _ in range(num_episodes):
policy = lambda env, obs: env.action_space.sample()
done = False
obs = env.reset()
# cnt = 0
while not done:
action = policy(env, obs)
obs, _, done, info = env.step(action) # env.step
# cnt += 1
# print(cnt)
episodes.append(
info['episode']
) # if done is True, info stores the 'episode' information and 'episode' is written in a file(e.g. "~/planet/log_debug/00001/test_episodes").
return episodes