in support/retro_contest/agent.py [0:0]
def random_agent(env, *args):
env.reset()
while True:
action = env.action_space.sample()
try:
ob, reward, done, _ = env.step(action)
except gre.ResetError:
done = True
if done:
env.reset()