PEARLWorker

[braunjon]

Hi, I have two issues that I do not quite understand why they are part of the code

1. The code overwrites the action a in the deterministic case:
   

   garage/src/garage/torch/algos/pearl.py

   Lines 743 to 746 in b4abe07

   	a, agent_info = self.agent.get_action(self._prev_obs)
   	if self._deterministic:
   	a = agent_info['mean']
   	a, agent_info = self.agent.get_action(self._prev_obs)

   
   There is an open pull request about this here: Fix double action sampling in PEARLWorker #2275.

2. I was wondering if the context is ever used in self.agent. As far as I understand in the pearl.py file we never use the context of self._policy and it is also not used within the class ContextConditionedPolicy.
   

   garage/src/garage/torch/algos/pearl.py

   Lines 754 to 759 in b4abe07

   	if self._accum_context:
   	s = TimeStep.from_env_step(env_step=es,
   	last_observation=self._prev_obs,
   	agent_info=agent_info,
   	episode_info=self._episode_info)
   	self.agent.update_context(s)

Some hints are appreciated
Thanks