b72212719d89f6d9e6a2c92f22f6749470280ed5,catalyst/rl/utils.py,OffpolicyReplayBuffer,push_episode,#OffpolicyReplayBuffer#Any#,155

Before Change

        with self._store_lock:
            observations, actions, rewards, dones = episode
            episode_len = len(rewards)
            self.len = min(self.len + episode_len, self.capacity)

            indices = np.arange(
                self.pointer, self.pointer + episode_len
            ) % self.capacity

After Change

            self.dones[self._pointer:self._pointer + episode_len] = \
                np.array(dones)

            self._trajectories_lens.append(episode_len)
            self._pointer = self._pointer + episode_len

    def recalculate_index(self):
        with self._store_lock:

Instances