5ec418e8146079d6797bb15b0f5e777862815df6,env.py,Env,reset,#Env#,54

Before Change

  def reset(self):
    observation = self._reset()
    // Press fire (raw action 1) to reset game if needed
    if 1 in self.actions.keys():
      observation, _, done = self.step(1)
      if done:
        self._reset()
      observation, _, done = self.step(2)
      if done:
        self._reset()
    return observation
 
  def step(self, action):
    // Repeat action 4 times, max pool over last 2 frames

After Change

    observation = self._get_state()
    self.state_buffer.append(observation)
    self.lives = self.ale.lives()
    return torch.stack(list(self.state_buffer), 0)

  def step(self, action):
    // Repeat action 4 times, max pool over last 2 frames
    frame_buffer = torch.zeros(2, 84, 84, device=self.device)

Instances