Improve normalizer for scalar

2026-09-12 12:05:35 +08:00 · 2017-10-07 21:23:02 -06:00
parent fae9a85f31
commit 51f125c01e
3 changed files with 13 additions and 4 deletions
@@ -79,7 +79,7 @@ class ProximalPolicyOptimization:
                batched_steps += 1
                episode_length += 1

-                reward = np.asscalar(self.reward_normalizer(np.array([reward])))
+                reward = self.reward_normalizer(reward)
                rewards.append(reward)

                if done: