def pop(self, warn=True, peek_d=None):
reward_d = {}
done_d = {}
info_d = {}
err_d = self.pop_errors()
for i, reward_buffer in self.reward_buffers.items():
name = self.names_by_id[i]
reward, done, info = reward_buffer.pop(peek_d.get(name))
reward_d[name] = reward
done_d[name] = done
info_d[name] = info
# TODO: use FPS here rather than 60
if warn and any(
info.get('stats.reward.count', 0) > 60
for info in info_d.values()):
logger.warn(
'WARNING: returning more than 60 aggregated rewards: %s. Either your agent is not keeping up with the framerate, or you should have called ".reset()" to clear pending rewards and reset the environments to a known state.',
{
name: '{} (episode_id={})'.format(
info['stats.reward.count'],
info.get('env_status.episode_id'))
for name, info in info_d.items()
})
return reward_d, done_d, info_d, err_d
def pop(self, warn=True, peek_d=None):
reward_d = {}
done_d = {}
info_d = {}
err_d = self.pop_errors()
for i, reward_buffer in self.reward_buffers.items():
name = self.names_by_id[i]
reward, done, info = reward_buffer.pop(peek_d.get(name))
reward_d[name] = reward
done_d[name] = done
info_d[name] = info
# TODO: use FPS here rather than 60
if warn and any(info.get('stats.reward.count', 0) > 60 for info in info_d.values()):
logger.warn('WARNING: returning more than 60 aggregated rewards: %s. Either your agent is not keeping up with the framerate, or you should have called ".reset()" to clear pending rewards and reset the environments to a known state.',
{name: '{} (episode_id={})'.format(info['stats.reward.count'], info.get('env_status.episode_id')) for name, info in info_d.items()})
return reward_d, done_d, info_d, err_d