From 564d5f5a8140a387d1c19c30724f9f288411993f Mon Sep 17 00:00:00 2001 From: "K.R. Zentner" Date: Sat, 8 May 2021 14:06:21 -0700 Subject: [PATCH] Fix double action sampling in PEARLWorker --- src/garage/torch/algos/pearl.py | 1 - 1 file changed, 1 deletion(-) diff --git a/src/garage/torch/algos/pearl.py b/src/garage/torch/algos/pearl.py index a5301aa4d..ff5eb880b 100644 --- a/src/garage/torch/algos/pearl.py +++ b/src/garage/torch/algos/pearl.py @@ -743,7 +743,6 @@ def step_episode(self): 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) es = self.env.step(a) self._observations.append(self._prev_obs) self._env_steps.append(es)